From first sign-in to production agents

Five minutes. Know what each item in the sidebar is for before you build anything.

The top of the sidebar shows your Working organization. The bottom has Log out and Minimize sidebar. Your subscription and SSO live on the Account page, reached from Settings.

You build, run, schedule, and check the cost of a real agent: the Competitor Watch. It needs no connections, so you can finish it on a brand-new instance in about 20 minutes.

Part 1 - Design it with the Agent Builder (5 minutes)

1. In the sidebar, open AI Agents → Agent Builder. You start on step 1, Describe.
2. In Your role, type Head of Marketing.
3. In Describe the business need, paste:

   Every week, research our three main competitors on the public web: Acme Analytics,
   Northwind Data and Contoso Insights. For each one, find what changed recently:
   new products or features, pricing changes, press releases or news, and notable hiring.
   Read the actual pages rather than guessing from search snippets. Save the findings as
   a spreadsheet with one row per finding (competitor, category, finding, source URL, date),
   then give me a short summary of what it means for us and the three things we should
   consider doing.

   Replace the three names with your real competitors if you like.
4. Press Design my agent. The frontier builder model takes 20 to 60 seconds.
5. You are now on step 2, Review. Check three things:
   • What it will do lists a numbered plan that searches, reads pages, saves a CSV, and summarizes.
   • Capabilities it will use shows Web: search, Web: read a page, and Files: save a CSV report.
   • Not covered is empty or names only things the web cannot answer, such as private pricing.
   Open View the agent's instructions and find the seven prompt parts from Module 4. Then press Continue.
6. Step 3, Connect, should say nothing is missing, because web research needs no connection. Press Continue.
7. Step 4, Create: set Agent name to Competitor Watch, keep the description, and press Create agent.

Part 2 - Inspect and tune it in the editor (5 minutes)

1. Open AI Agents → Agents. Find Competitor Watch under Your Agents and press Edit.
2. Check Tools & Capabilities: Web Search is ticked. The page-reading and CSV tools the builder chose have no checkbox in the editor, and saving keeps them. Leave everything else unticked.
3. Set Temperature to 0.3, because research should be consistent rather than creative.
4. Make sure Run Prompt holds one concrete instruction, for example: Run this week's competitor research and deliver the spreadsheet and summary.
5. Try an AI edit. In Describe the agent you need at the top of the editor, type Add a "threat level" column (low, medium, high) to the spreadsheet. Optionally press Attach files or paste a screenshot (Ctrl+V) of a report layout you like; up to 10 files are accepted. Press Generate with AI, review the changed fields, then press Save Agent.
6. Press Versions on the card. You will see the original and your edit, and you can roll back with Publish This Version.

Part 3 - Run it and read the result (5 minutes)

1. On the agent card, press Run. Weavr confirms with Agent started! and the run prompt executes in the background.
2. Open AI Agents → Monitor. While it works, the agent shows as running and the Executing now tile counts it.
3. When it finishes, click the Competitor Watch row. Under Run traces - what each run was asked and did, click the newest run. You see each search and page read in order, what came back, and at the bottom the Final answer: the per-competitor findings, a download link for the CSV (valid for 24 hours), and the "what it means for us" summary.
4. Back on the agent's row, check Cost this month. That is what this agent has spent on AI so far this month.
5. Quality check: every finding should carry a real source URL. If one does not, edit the system prompt's data rules to say "every row must cite the page it came from" and run again.

Part 4 - Schedule and govern it (5 minutes)

1. Press Edit. Under Schedule / Automation, choose ⏰ Run on Interval and set Run Every (minutes) to 10080 (one week). Press Save Agent.
2. The card now shows the schedule with pause (⏸️) and resume (▶️) controls. Schedules survive restarts and upgrades.
3. Ask an Org Admin to do this step if you are not one. Open Settings → Usage & Cost Controls. In Per-Agent Limits & Usage (today / month), give Competitor Watch a daily cost limit of 1 and save. A runaway research loop can then never cost more than $1 a day.

Done when

• A manual run returns findings with source URLs, a working CSV link, and a three-point recommendation.
• The weekly schedule is visible on the card, and Monitor shows a non-zero Cost (month) for the agent.
• You can name the build cost (the design) and the run cost (the run) on the Dashboard.

You chain your Competitor Watch into a two-step Competitor Brief workflow that writes an executive brief, delivers it as a PDF, can be started from any system, and shows its own monthly cost. About 25 minutes. Finish Guided build 1 first.

Part 1 - Create the second agent (5 minutes)

1. Open Agents and press Create Agent.
2. In Describe the agent you need, type:

   An executive brief writer. It receives competitor research findings and turns them
   into a one-page brief for the leadership team: a three-sentence headline summary,
   a table of the most important changes per competitor with a threat level, and three
   recommended actions with an owner role and a rough effort for each. It uses only the
   findings it is given and never invents facts.

3. Press Generate with AI. Check that Tools & Capabilities is empty, because this agent only writes. Set Agent Name * to Executive Brief Writer and press Save Agent.

Part 2 - Lay out the flow on the canvas (5 minutes)

1. Open AI Agents → Workflow and press New. Type Competitor Brief in Name this flow.
2. From Your agents on the left, drag Competitor Watch onto the canvas, then drag Executive Brief Writer below it. Clicking an agent also adds it.
3. Connect them: drag from the green dot on Competitor Watch to the blue dot on Executive Brief Writer. An arrow appears.
4. Click the amber ✎ describe this handoff label on the arrow. In What does Competitor Watch hand to Executive Brief Writer - and what should it do with it?, type:

   Hand over every finding (competitor, category, finding, source URL, threat level)
   and the CSV link. Write the one-page leadership brief from these findings only, and
   include the CSV link at the end as the evidence.

   Press Done, then ⤓ Arrange to tidy the layout, and Save flow.

Part 3 - Decide how it starts and where results go (5 minutes)

1. In How this flow starts, keep Manually, from this page ticked and also tick Any other system (webhook). Because the flow is saved, a trigger example appears. Press Copy the trigger example and keep it for Part 5.
2. In Deliver the results, tick PDF report. Leave the push destination as Don't push anywhere for now. Later you could push the brief into a CRM or a mailbox here.
3. Optional: under Reference documents, press 📎 Attach documents and add your positioning or brand-voice document. Every step receives it on every run.
4. Press Save flow again so these choices are stored with the flow.

Part 4 - Run it and read the results (5 minutes)

1. In Run this flow, type the client ask: This week's competitor brief for the leadership meeting. Press ▶ Run.
2. The run is queued and executes in the background. You can leave the page. Run queue & history shows it as queued, then running, then completed.
3. Click the run to open Results: each step's output in order, and the PDF to download.
4. Open Monitor. Competitor Brief appears as a workflow row. Its Cost (month) is the AI cost of every agent call made during its runs. Click the row to see its success rate, average duration, and cost this month.

Part 5 - Start it from outside, and try a controller (5 minutes)

1. Run the trigger example you copied from any terminal, or from a tool like Zapier. It returns at once with a run id; the run appears in Run queue & history. API calls need an organization API key; see Module 9.
2. Optional, for bigger flows: in Workflow Controller, tick Let a controller orchestrate this flow and write its job description, for example Have Competitor Watch research first, then have the Executive Brief Writer write the brief; if research finds nothing new, say so in one paragraph instead. The controller then decides the order per request and writes the final answer. Choose its Manager model, the busiest and most billed model in the flow.

Done when

• A manual run completes both steps, and the PDF brief cites only the research findings.
• A webhook call starts a run that appears in the queue.
• Monitor shows the workflow with its own Cost (month).

Outcome: you can explain the platform's architecture, its vocabulary, and the build-versus-run model split to a colleague.

1.1 The mental model: ETL for AI

Weavr treats every system in your business as a live data stream. An agent extracts context from those systems through its tools, transforms it with AI reasoning, and loads actions back: a draft reply in a mailbox, a note in the CRM, a CSV report, a workflow hand-off. If a system has an API, Weavr can put it to work.

1.2 Your instance

Every organization runs on its own dedicated, single-tenant instance. Your agents, workflows, connections, credentials, and usage data live only there. Managed plans run on Powdr-held provider keys with automatic failover across Gemini, Groq, Anthropic, and OpenAI, so a single provider outage never stops a workflow. Enterprise Max instances bring their own keys instead.

1.3 The build-versus-run split

Weavr uses the most capable model available to design things and a cost-efficient model to run them. When the Agent Builder designs an agent, a frontier model (Anthropic Opus by default) writes the instructions and picks the tools. The agent you save then executes on the platform default (Gemini). You will see this on the Review step as "Designed by ... will run on ...". The same split appears in your usage dashboard as build cost versus run cost.

1.4 Vocabulary you will use every day

Outcome: you can sign in, invite teammates with the right roles, block access, and enable company SSO.

2.1 Signing in

1. Go to your instance's login page and enter your work email (personal domains like gmail.com are rejected).
2. If your company has SSO enabled, the button becomes Continue with Microsoft (or Google). Otherwise you receive a six-digit one-time code by email. No passwords exist anywhere in Weavr.
3. First user on a company domain becomes the organization admin; later users on the same domain join automatically as operators.

2.2 The four organization roles

2.3 Managing users (Org Admins)

1. Open Settings → User Management.
2. Invite: enter a work email, pick a role, press Send Invite. They receive a sign-in link and join with that role on first login, even from a domain you have not claimed.
3. Change a role: use the dropdown on the member's row.
4. Block: reversibly cuts a member's access (they keep their history); Remove is permanent. Nobody can block or demote the organization creator.

2.4 Enabling SSO (Org Admins)

1. In your identity provider (Microsoft Entra or Google Workspace), create an app registration with the redirect URI shown on your Account page (it ends in /api/v1/auth/sso/callback).
2. On the Account page, open the SSO card and enter the client ID, client secret, and (for Microsoft) the directory tenant ID, then enable it.
3. On the Account page, use Add company domain for every email domain your staff sign in with, including alias domains, so nobody is turned away at the door.

Outcome: you can connect a supported system, verify it works, attach it to agents, and stand up a connection to a system Weavr has never seen.

3.1 What a connection is

A connection is one system's credentials, stored encrypted on your instance and shared by every agent you attach it to. Tools only work when their connection exists and is attached to the agent. The catalog covers:

3.2 Connecting a system

1. Open Connections. Use the Connect a system card, or pick a tile under Available platforms.
2. Choose the system in Pick the system you want Weavr to work with. Weavr asks only for the fields that system needs and explains where to find them. Microsoft systems need a tenant ID, client ID, and client secret from an Entra app registration. Where a system offers it, I have a personal access token instead is the faster route.
3. Press Test & connect. Weavr makes a real read-only request. A failure names the step that failed; fix it before you continue.
4. The connection appears under Your connections, with Test and Remove. Its tile under Available platforms shows Added.
5. Attach it to an agent in the editor under Connect your systems → Where should it get information?, or let the Agent Builder's Connect step do it for you.

3.3 A system Weavr has never seen

Type the platform's name in Connect a system. If it is not in the catalog, the Connector Designer designs a connector from that system's own API: the endpoints, the authentication style, the required fields, and a real read-only test request. You review the design, add credentials, and publish. The definition becomes available to your whole organization.

3.4 Health and hygiene

• When an agent reports "connection missing" or a tool silently does nothing, check that the agent has the connection attached and that Test still passes.
• Rotated a secret in the source system? Update the connection immediately; agents fail with a clear credential error otherwise.
• Two connections to the same system with different credentials confuse everyone. Keep one per system unless an agent needs a separate, narrower key.
• Remove connections nothing uses. Fewer live credentials is always better.

Outcome: you can take a business need from plain English to a saved, reviewed agent, change it safely with AI, and you know what every field in the editor does.

4.1 The guided Agent Builder (recommended path)

Open Agent Builder in the sidebar, or Guided Builder on the Agents page. The wizard has four steps:

1. Describe. Fill Your role and Describe the business need the way you would to a colleague, naming the systems your data lives in. You can attach files, add reference links, or tap the mic and dictate. Need a starting point? offers examples. Press Design my agent.
2. Review. The frontier model returns a full design: What it will do (a numbered plan of the tools it will call, in order), Capabilities it will use, and an honest Not covered list. Open View the agent's instructions to read and edit the prompt. Nothing is saved yet.
3. Connect. Shows which required systems are connected and which still need credentials, with a set-up action for each gap. Re-check after connecting. You can continue with gaps; the agent will say what it cannot reach rather than guess.
4. Create. Confirm Agent name and Description and press Create agent. It appears on the Agents page, set to run on the cost-efficient default model.

4.2 Generate and change agents with AI, with attachments

On the Agents page, Create Agent opens the editor with Describe the agent you need at the top. Describe the job and press Generate with AI; the form fills in for your review, including tools, connections, and a schedule when the work is clearly recurring. You still press Save Agent yourself.

The same box changes an existing agent: open it with Edit, describe the change ("add a risk section", "only look at the last 30 days"), and press Generate with AI. Every AI change is recorded in the agent's Evolution history.

• Show, don't just tell. Press Attach files, drag files onto the box, or paste screenshots with Ctrl+V. Up to 10 files per request: images (PNG, JPEG, GIF, WebP), PDF, Word (.docx), and text, Markdown, CSV, or JSON. The counter shows how many you have added.
• Good attachments: a screenshot of a report layout you want copied, a sample of the source data, the process document the agent should follow, or an example of output that was wrong.
• Limits: 10 MB per file and 30 MB per request. The files inform this one design; the history keeps only their names.

Quick Start Templates at the top of the Agents page create a ready-made agent in one click, which you then adjust.

4.3 Anatomy of an agent

4.4 Writing a system prompt like the platform does

Every strong Weavr agent prompt has seven parts. The builder writes these for you; when you edit by hand, keep all seven:

1. Persona: the professional role and who it reports to.
2. Objective: the business question a successful run answers.
3. Numbered execution plan: each tool by exact name, in order, and what feeds what.
4. Data rules: report real retrieved data only; never invent or estimate figures; say plainly when a source returns nothing.
5. Output format: the exact sections of the final report, in order, so every run is consistent.
6. Edge cases: what to do when a connection is missing, a tool errors, or data is empty. Say so and continue; never guess.
7. Tone: concise, professional, decision-ready.

4.5 Versions, history, and safekeeping

• Every save creates a version. Versions on the card lists them; View Details shows one and Publish This Version rolls back. Experimenting with prompts is safe.
• History on the card opens the agent's activity log: total, successful, and failed actions and recent activity.
• Export downloads your organization's agents and workflows as one file, never containing credentials. Import loads such a file, for example into another instance. Back Up sends a copy to GitHub, OneDrive or SharePoint, or Google Drive.

Outcome: you can run agents on demand and on schedules, stop and pause them, and know every way a request can reach an agent.

5.1 Ways to run an agent

• Run on the agent card executes the run prompt once, now, in the background. Stop halts it.
• Tasks in the sidebar gives an agent a one-off job: New Task, pick the Agent, a Task Type such as Chat / Query or Document Review, type the Input / Prompt, and press Run Task. The answer appears under the task's Response tab. This is the way to ask an agent a question directly.
• Schedules run it on their own (5.2).
• A workflow runs it as one of its steps (Module 6).
• Slack or Teams: a bound channel sends every message to the agent and posts the answer back (Module 10).
• The API: other software can ask an agent to do one thing and get the answer (Module 9).

5.2 Schedules

In the editor, Schedule / Automation offers four choices:

Scheduled agents show their schedule on the card with pause (⏸️) and resume (▶️) controls. Schedules survive restarts and upgrades; an active schedule resumes on its own. Tick Start schedule immediately when saving if the first run should happen now rather than at the first interval.

5.3 Reading results

An agent's answer and every step it took are in its run trace on the Monitor page (Module 7). Agents that deliver files return a download link in their answer, and agents with a destination push the result into that system.

Outcome: you can design a workflow by description or by hand, decide how it starts and where its results go, put a controller in charge when the order varies, and run it in the background.

6.1 Two ways to build

• Guided Builder (toolbar, top left). Give your role, describe the process, and optionally attach reference documents. Press Design my workflow. Weavr proposes the workflow name, the steps, the agents, the connections it needs, and an example client ask. Press Build this workflow, read the Build review (what was created versus what you provided), then Done - show the canvas. Missing agents are created for you.
• By hand on the canvas. Drag agents from Your agents on the left, or click one to add it. Connect steps by dragging from a step's green dot to the next step's blue dot. ⤓ Arrange lays the flow out top-down; Fit and 1:1 control the zoom. Name the flow and press Save flow.

A saved flow can be changed later with Update with builder: describe the change and the builder redesigns it in place. Every design revision is kept under Versions, so you can roll back. Open saved flows from My saved flows….

6.2 The Client Ask and hand-offs

Every starting step (one with no incoming arrow) receives the Client Ask word for word. A downstream step receives the original ask, the upstream step's output, and the hand-off written on the arrow. An arrow with no hand-off shows an amber ✎ describe this handoff label; click it and answer What does A hand to B - and what should it do with it? Steps that do not depend on each other run in parallel.

A hand-off is the contract between two agents. Name what arrives and what must leave: "Hand over the verified contacts (name, company, email). Draft a personalized outreach email for each one." Vague hand-offs such as "continue" produce vague results.

Click the pencil (✎) on a step to edit that agent in place: its label, instructions, run prompt, and model, or describe a change and press Improve.

6.3 How a flow starts

Under How this flow starts, tick every way the flow may be called. Each card records its setup with the flow:

POST /api/v1/agents/workflows/{workflow_id}/trigger
{ "message": "the ask", "source": "your-system", "context": { ... } }

The call returns at once with a run id and a status URL; the run joins the queue.

6.4 Delivering results

Deliver the results adds downloadable files (HTML report, PDF report, Spreadsheet (CSV)) and can push the result into a connected system with an instruction, for example "Log a note with the summary on each account mentioned". Results always appear in Weavr as well.

6.5 The Workflow Controller

A fixed step order suits processes that never vary. When the order depends on the request, tick Let a controller orchestrate this flow and write the controller's job description. A manager agent then reads each request, dispatches work to the flow's agents in whatever order the job needs, and writes the final answer.

• Manager model: the controller's model. It reads every request and writes every reply, so it is usually the busiest and most billed model in the flow.
• Agents' model in this flow: the model every member agent runs on when this flow calls it. Agents are shared between flows, so this is set per flow.

6.6 Reference documents

Brand voice rules, company context, or process documents attached under Reference documents are saved with the flow and handed to every step, and the controller, on every run. PDF, Word, text, Markdown, CSV, or images; up to 8 files of 10 MB.

6.7 Background runs, results, and cost

Runs execute in the background, so you can leave the page. Run queue & history shows each run as queued, running, completed, or failed; click one to reopen its Results with every step's output and the delivered files. Run records are kept for 90 days and full results for 30. Monitor shows each workflow's success rate, average duration, and Cost (month), which covers every AI call made during its runs (Module 7).

Outcome: you can read the Monitor page in a minute, find any run and its full trace, and say what each agent and workflow has cost this month.

7.1 The health strip

Open AI Agents → Monitor. Pick a window with Last hour, 24 hours, or 7 days; every figure on the page follows it. Live pauses or resumes the five-second refresh. A Needs attention banner appears at the top when something is failing.

7.2 Agents & workflows

The fleet table lists every agent and workflow you can see. Filter with All, Agents, Workflows, or Problems, or use Search…. Running items sort first, then problems, then the most recently active.

• Status: running (working this second), idle (on and waiting), stopped (switched off), or error (the last thing it did failed).
• Schedule: an agent's schedule; for a workflow, its step count or controller.
• Runs and Success: in the selected window. A red number next to runs counts failures.
• Cost (month): AI spend this calendar month. For an agent, every model call it made. For a workflow, every model call made during its runs: the controller plus each member agent. A member agent's cost therefore appears both on its own row and inside each workflow that used it, so do not add the two columns together.

7.3 Details and run traces

Click a row to open its detail panel. An agent shows its status, schedule, lifetime actions, Cost this month, 24-hour activity, what it does most, and its recent runs. A workflow shows its shape, last run, average duration, lifetime runs, success rate, and Cost this month. Click a run for its model, tokens, cost, and duration, then View run trace - every step: each model turn, every tool call with its arguments and what came back, any errors, and the Final answer. Steps that change data in another system are marked changes data. After fixing a broken agent, Reset this agent's stats clears its old failure counts.

7.4 Activity

The Activity stream on the right is a live feed of what the platform is doing, filterable to Errors or Warnings. Errors carry a plain-language cause and fix. Events reported from other systems through intake (Module 9) land here too.

7.5 Diagnosing failures

Outcome: you can say where every AI dollar went, by agent, by workflow, and by model, and set guardrails so spend never surprises you.

8.1 Where to see spend

A workflow's cost covers every model call made during its runs: the controller and every member agent. A member agent's calls inside a workflow therefore count on the agent's row and on the workflow's row. Use the workflow figure to price a process and the agent figure to price a worker; do not add them together.

8.2 The guardrail stack

1. Plan ceiling: the monthly AI usage included in your plan. At the ceiling, model calls stop with a clear "Monthly usage ceiling reached" message, and nothing bills past it. Enterprise Max runs on your own provider keys and has no ceiling; its spend is your provider bill.
2. Platform limits (Settings → Usage & Cost Controls): Platform Daily Cost Cap and Platform Monthly Cost Cap lower your effective cap; they can never raise it above the plan.
3. Agent defaults: Default Agent Requests / Day and Default Agent Cost / Day ($) apply to every agent without its own limit.
4. Per-agent limits: requests per day, dollars per day, and an Enabled switch for each agent. Cap experimental agents low. A blocked call returns a clear "[Usage Limit]" error.

8.3 Build cost versus run cost

Every model call is metered and categorized. Build is creation work: the Agent Builder, the Connector Designer, and presentation and project-plan generation. Run is agents and workflows doing their jobs, including every intermediate step in a tool loop and failed provider attempts, which still consume input tokens. "Why was this month expensive" always has an answer: you built a lot, or you ran a lot, and Monitor tells you which agent or workflow.

8.4 Plans and the trial

• Monthly plans start with a free trial that includes a small AI allowance; the card on file converts automatically unless cancelled. Your plan, status, and ceiling are on the Account page.
• Enterprise Max bills annually, uses your own provider keys entered under Settings → AI Providers, and has no ceiling.

Outcome: you can give another system a key, have it run an agent or workflow and get the answer, and pipe events from any platform into Weavr.

9.1 API keys

Software proves which organization it acts for with an organization API key. Keys start with wvk_, are shown once when created, and are stored only as a hash. Send one as X-Weavr-Key: wvk_… or Authorization: Bearer wvk_….

• Org Admins create keys with POST /api/v1/api-keys/, or let the Slack and Teams guided setup create one.
• A key has one of two roles: operator (run agents and workflows, read results) or agent_builder (also create and change agents). Give each system its own key with the smallest role it needs, so you can revoke one without breaking the others.

9.2 Running agents and workflows from other software

The External Agent API is the way other systems put Weavr to work:

POST /api/v1/external/agents/{agent_id}/invoke
{ "message": "Summarize this account's open issues", "context": { ... },
  "source": "crm", "timeout_seconds": 120 }

• Wait mode: the call waits up to timeout_seconds and returns the answer. If the run is still going, it returns a run id; poll GET /api/v1/external/runs/{run_id} until it reports completed or failed. A not-found answer while the run is still working means keep waiting.
• Callback mode: pass callback_url and the call returns at once; the result is POSTed to your URL when ready.
• Workflows work the same way with /api/v1/external/workflows/{id}/invoke. GET /api/v1/external/agents and /workflows list what the key can reach.

The workflow trigger URL from the Workflow page (POST /api/v1/agents/workflows/{id}/trigger with message and optional context) starts a background run and returns a run id and status URL.

9.3 Event intake

Any system that can make an HTTP call can report into your instance:

POST /api/v1/intake/{source}/events
{ "events": [ { "level": "error", "message": "...", "timestamp": "..." } ],
  "instance_id": "optional" }

• Error events land in the Monitor activity stream with a diagnosed cause and fix.
• Identical errors repeating batch after batch are muted for six hours and surface once with a repeat count, so a stuck upstream cannot flood your feed.
• Escalation: name an agent "<source> support", "<source> intake", or "<source> monitor" (for example Kyni Support for source kyni) and give it a Teams or Outlook connection. When a batch brings three or more fresh errors from that source, Weavr hands the batch to that agent to triage, group, and notify, once per mute window.

9.4 The reference

API Docs in the sidebar is the full interactive reference: everything the UI does, the API does.

Outcome: your team asks an agent for work from the channel they already live in, and the answer comes back to that channel - drafts as attachments, long jobs with progress updates.

10.1 Slack - about a minute

Weavr runs one Slack app for the whole platform, so you install it rather than build anything. There is no app to create, no token to copy, and no secret to paste. Every organization installs that same app into its own Slack workspace and gets its own, isolated connection - your workspace can only ever be attached to your organization.

1. Install. Connections → choose Slack → Add to Slack. Slack shows one approval screen listing what the app can do; click Allow. You land back in Weavr, connected. The install is tied to your organization - it can never attach to anyone else's, and a workspace already connected to another Weavr organization is refused rather than taken over.
2. Requirements. Your Weavr subscription must be active - the install is refused otherwise, with the reason. Starting the install from Slack's own directory does not work by design: only a link that begins inside Weavr knows which organization you are.
3. Bind a channel. A channel answers only when it is bound to an agent or a workflow - an unbound channel says so rather than guessing. Create a channel (or use an existing one), then bind it from the guided setup, or via the API:

   POST /api/v1/slack/bindings
   { "team_id": "T…", "channel_id": "C…", "channel_name": "nimbus",
     "target_type": "workflow", "target_id": "<workflow id>" }

   Use "channel_id": "*" to set a workspace-wide default that also covers DMs.
4. Use it. In a bound channel, just type - every message goes to its agent, no mention needed. Elsewhere, mention the app - @Weavr draft a blog post on lead times - or DM it. If the workspace's app defines a slash command, that works too: the command name belongs to whatever the connection is defined as, not to Weavr itself, so each organization's app can name its own. You get an acknowledgement immediately, a progress note every 30 seconds on long jobs, and the answer when it lands. Replies in the same thread continue the conversation: if the agent asks a clarifying question, just answer it - no need to mention the app again. The same thread is where you give feedback. Reply with what you want different - shorter, add pricing, make it about Q3 - and the agent revises the piece and posts the updated version, keeping everything you didn't mention as it was. Keep going until it's right; that back-and-forth is the point, not a one-shot answer.
5. DMs pick their handler. A DM is its own channel, so it routes in order of how explicit you are: start the message with a bound channel's name - #nimbus draft a launch post - to use that channel's agent; once a #channel prefix has picked an agent, plain replies in that DM keep going to it - to whatever that channel is bound to right now - so answering an agent's clarifying question continues the work instead of starting over. Naming another bound channel switches; after two quiet hours the DM resolves the normal way again. Otherwise the workspace-wide * default answers if one is bound; and when the workspace has exactly one bound channel, a plain DM goes there automatically. A DM that can't resolve replies with the bound channels to choose from - never a dead end.

• Long answers arrive as a markdown attachment named after the piece, with a short summary in the message - not as a wall of chat.
• If Weavr restarts mid-run, the answer is delivered late or you are told plainly the request was interrupted. Silence is never the outcome.

10.2 Microsoft Teams - about five minutes, no administrator needed

Teams has no one-click install like Slack's. A conversational Teams bot normally needs an Azure Bot registration and admin consent - so Weavr uses Power Automate instead, which you build from inside Teams yourself. The flow does the talking: it catches channel messages, sends them to Weavr, waits for the answer, and posts it back.

1. Create an API key. The flow proves which organization it belongs to with a key. The guided setup creates one for you (shown once - paste it straight into the flow), or an org admin can mint one at POST /api/v1/api-keys/.
2. Pick the agent. Still in the guided setup, bind a channel name (or * as the default for every channel) to the agent or workflow that should answer. The flow you build next carries that name - you never look up an agent id, and you can change which agent answers later without touching the flow.
3. Start the flow. In the Teams channel: … menu → Workflows → create from blank. Trigger: When a new channel message is added, pointed at your team and channel.
4. Add the HTTP action. The guided setup generates this block exactly - copy it rather than typing. In outline: POST to /api/v1/teams/message with headers Content-Type: application/json and X-Weavr-Key: <your key>, and a body carrying the message text, the conversation id, and the channel name you bound in the previous step.
5. Post the answer back. Add Post message in a chat or channel, pick the same channel, and set the message to the HTTP action's reply field. Weavr answers synchronously, so the flow owns the whole round trip - there is nothing else to wire.
6. Try it. Post in the channel. If nothing comes back, Power Automate's run history shows each step's input and output - the failing step is highlighted, and the guided setup can read the error back to you in plain language.

Outcome: you know which ready-made agents ship with Weavr, what each needs before it can run, and how the vCISO and vCFO services are driven.

11.1 What ships with Weavr

Built-in agents appear on the Agents page like your own. Weavr keeps their instructions up to date with each release; your schedules and connections on them are kept.

The Powdr hub also runs internal agents that tenants do not see: the AP Invoice Intake Agent, the Agent Reliability Supervisor (a daily review of every run trace and error, emailed as a report), and Powdr's own vCFO.

Quick Start Templates on the Agents page add more starting points in one click, including Sales Prospecting, Email Virtual Assistant, IT Support, Documentation Writer, Code Reviewer, and Kyni Support agents.

11.2 Running the vCISO for a client

1. Connect the client's tenant. For Microsoft 365, use the M365 & Azure Tenant Audit connection: the client's administrator grants admin consent to the read-only audit app for their tenant. For Google Workspace, use the Google Workspace Security Audit connection: a Google Cloud service account with domain-wide delegation for the read-only audit scopes, and the client's admin listed as admin@client.com = Client Name.
2. Allowlist the client. The vCISO reviews only tenants and domains on its allowlist, so it can never be pointed at an organization you have not onboarded. Managed-service teams usually do this from Kyni (below), which calls /api/v1/vciso/targets with an agent_builder key.
3. Run the review. Ask the vCISO by client name, from Kyni, from Tasks, or with the External API. A review takes several minutes and returns structured findings: severity, evidence, affected items, the recommended fix, and the effort.

Setting up Google Workspace reviews (once per organization)

Google Workspace reviews use one Google Cloud service account that each client authorizes for read-only access - the Google counterpart of the Microsoft audit app. It takes about five minutes, then a few minutes per client.

1. Create a Google Cloud project. At console.cloud.google.com, open the project picker (top left) → New Project, name it (for example vCISO Audit), create it and select it.
2. Turn on the Admin SDK API. Menu → APIs & Services → Library → search Admin SDK API → Enable.
3. Create the service account. Menu → IAM & Admin → Service Accounts → Create service account, name it (for example vciso-workspace-audit) and click Done. It needs no roles.
4. Create its key. Open the service account → Keys → Add key → Create new key → JSON. A .json file downloads - it is a password: keep it private and delete it after the next step. If Google says key creation is disabled, your Google organization blocks keys; an Organization Policy Administrator must allow it for this project (constraint iam.disableServiceAccountKeyCreation). Prefer a terminal? In Cloud Shell the same steps are: gcloud projects create vciso-audit-<yourcompany> && gcloud config set project vciso-audit-<yourcompany> && gcloud services enable admin.googleapis.com && gcloud iam service-accounts create vciso-workspace-audit, wait 30 seconds, then gcloud iam service-accounts keys create key.json --iam-account=vciso-workspace-audit@vciso-audit-<yourcompany>.iam.gserviceaccount.com && cloudshell download key.json.
5. Add the connection in Weavr. Connections → Connect a system → search google workspace → Google Workspace Security Audit. Paste the whole key file into Service Account Json, leave Delegated Admins empty and click Test & connect. The test confirms the key works.
6. Onboard each client from Kyni. MSP Operations → vCISO → the client → Onboard → Google Workspace, and enter the client's super admin email. Kyni adds the client to the connection and shows the service account's client ID and scopes.
7. The client authorizes read-only access. Their super admin opens admin.google.com → Security → Access and data control → API controls → Manage Domain Wide Delegation → Add new, enters the client ID (the client_id in the key file) and pastes these scopes, comma-separated, then clicks Authorize: https://www.googleapis.com/auth/admin.directory.user.readonly,https://www.googleapis.com/auth/admin.directory.domain.readonly,https://www.googleapis.com/auth/admin.directory.customer.readonly,https://www.googleapis.com/auth/admin.directory.rolemanagement.readonly,https://www.googleapis.com/auth/admin.directory.device.mobile.readonly,https://www.googleapis.com/auth/admin.reports.audit.readonly. All six are read-only; reviews work as soon as it is saved.

11.3 The vCFO

Each organization can have its own vCFO, bound to that organization's Kyni. It audits the books every day (cash, receivables, payables, the ledger, and integrity checks), answers finance questions, and returns a list of recommended actions. It is connected and driven from Kyni's Finance → vCFO page, which issues it a read-only finance key; its recommendations appear on the Kyni Finance Overview. Reconnecting issues a fresh key and replaces the old one in Weavr.

Outcome: you can find and change every instance-level setting, and you know which pages are for admins only.

12.1 Dashboard

The landing page. Tiles show Active Agents, AI Requests Today, AI Cost This Month, Team Members, and AI Models. Below them, Agent Metrics (this month) and Cost by Provider & Model (this month) with the build and run split, recent tasks, available templates, and quick actions to create an agent or task.

12.2 Models

Models lists every model your instance can use. Filter by provider, capability (chat, code, reasoning, analysis, vision), and status. Agents run on the platform default unless you choose otherwise in the editor or in a workflow's model settings.

12.3 Settings

12.4 Account

The Subscription card shows your plan, status, members, and usage against the ceiling, with the Billing portal. Org Admins also see Members with Add company domain, the Single Sign-On configuration, and, at the bottom, archiving (cancelling) the organization.

12.5 Other admin pages

• Workspaces (from the Dashboard): register a local directory, Git repository, S3 bucket, or email inbox that agents and tasks can work in, then Sync or Delete it.
• Deploy: the deployment manager, with Deploy and Rollback for each target and the deployment history. Ops Admin.
• Spoke Tenants and Pricing: platform administrators only. Every tenant instance, its spend with the build/run split, provider billing actuals, promo codes, and plan pricing.

Build the flagship sales agent end to end and prove every module at once. Allow 45 minutes.

Step 1: connections (Module 3)

Confirm your CRM connection tests green. If your CRM is unsupported, use Quick Connect to design the connector first. If you also bill from an ERP like Odoo, connect it too so revenue history is available.

Step 2: design (Module 4)

Open Agent Builder and describe: "I'm the director of sales. I want an agent that pulls our sales and lead pipeline from our CRM along with the revenue history for each client, then gives me a summary of the pipeline plus the next best action for every client, based on an engagement score built from active projects and services and how that correlates to average invoice size, so I can see which accounts we can grow and which are starting to slow down."

On Review, verify the plan: pipeline retrieval, portfolio retrieval, revenue summaries, engagement scoring, a CSV report, and a structured executive summary. Verify the instructions carry all seven prompt parts and an honest Not-covered list. Continue through Connect and Create.

Step 3: run and read (Modules 5 and 7)

Press Run. In Monitor, follow the tool rounds and confirm the run trace's final answer has the exact sections the prompt promised, with a downloadable CSV link. If the Kyni or CRM key is rejected, fix the connection and run again; the agent should have said plainly what it could not reach.

Step 4: schedule and govern (Modules 5 and 8)

Schedule it weekly (⏰ Run on Interval, 10080 minutes) and set a per-agent daily cost cap of $1. Check the Dashboard after the run: the design shows as build cost, the run as run cost. In Monitor, read the agent's Cost (month).

Step 5: compose (Module 6)

On the Workflow canvas, chain it into a second agent whose hand-off prompt produces a five-bullet summary for the leadership channel. Save the flow, tick Any other system (webhook), and copy its trigger example.

Pass criteria

• The agent runs green with a consistent, sectioned report and real (never invented) figures.
• Missing systems are stated plainly, not guessed around.
• The schedule survives a page reload; the workflow run appears in Run queue & history with both steps completed, and in Monitor with its own cost.

Build a mailbox-watching agent that drafts replies in a person's own voice and never sends. Allow 45 minutes.

Step 1: the mailbox connection

Connect Outlook with credentials that can read the target mailbox and create drafts (the Microsoft Graph application permissions Mail.Read and Mail.ReadWrite).

Step 2: learn the voice

Create the agent with only these tools: the sent-mail reading tool (outlook_list_sent_messages) plus inbox reading and the threaded draft tool (outlook_create_reply_draft). Do not grant outlook_send_email or outlook_reply_to_message; those really send. In a task (Tasks → New Task, Chat / Query), ask it to study the owner's sent mail in windows across recent months and produce a style profile: voice and tone, structure, characteristic phrases, response patterns, and verbatim example snippets. Paste the profile into the system prompt so the voice is fixed, not re-learned every run.

Step 3: the drafting workflow

Write the workflow section of the prompt with a strict tool budget: one client-list lookup, one inbox scan, one sent-mail snapshot to see what the owner already answered, then for each of at most five candidates: read the message, immediately create the threaded draft, move on. Skip newsletters, receipts, and anything already answered. End with a fixed-format run report. Above all: the agent never sends; drafts only. The draft tool also refuses to double-draft a thread that already has one, so polling is safe.

Step 4: schedule

Choose 📬 Monitor Inbox with Poll Every (seconds) set to 300. Confirm the first cycle in Monitor, and confirm a real draft appears threaded in the mailbox's Drafts folder for a qualifying email.

Pass criteria

• Drafts appear on the correct threads, in the learned voice, and nothing is ever sent.
• Repeated polls never produce duplicate drafts.
• Money or contract emails are flagged in the run report, not drafted.