Skip to main content

Operations with EnClaw — configure and use

A practical guide for the operations team, organised around the three things you asked for. Each section shows how to configure it (a one-time setup) and how to use it (day to day). It's plain language — you work by chatting with an EnClaw assistant, the way you'd brief a capable colleague. It pulls the numbers live from Productive; you stay in control and nothing is ever written back to Productive.

The three things, in your words:

  1. I want to know the profitability of our projects
  2. I want proactive flagging of low profitability
  3. I want to know our utilisation — e.g. what is our utilisation for FEDs over the next 4 weeks

How it works, in a nutshell

  • You ask in plain English — no reports to build, no spreadsheets to export.
  • It reads Productive live — projects, financials, and the forward booking schedule.
  • It's read-only — EnClaw never changes anything in Productive; it only reads.
  • It can watch for you — low-margin projects can be flagged automatically on a schedule, so you don't have to remember to check.

The setup basics (do this once)​

Setup is all menus and pasting — no coding. There are three one-time jobs: connect Productive, create an operations assistant, and (optionally) adjust a couple of settings.

1. Connect Productive (admin)​

This is what lets EnClaw read your data. You'll need a Productive API token and Organisation ID (in Productive: Settings → Organization → API integrations).

  1. Go to Settings → Integrations and add (or edit) the Productive integration.
  2. Paste the API token and Organization ID.
  3. Under Permissions, tick the read scopes the operations use cases need:
    • Read Projects
    • Read People
    • Read Time Entries
    • Read Bookings (utilisation)
    • Read Financials (profitability)
  4. Click Save, then click Test.
  5. Read the Permission probes result. Each scope shows a tick (✓ works), a warning (! needs attention), or a cross (✗ failed). You want ✓ against Read Bookings and Read Financials before relying on utilisation and profitability.

If the Read Financials probe shows a warning saying the endpoint wasn't found, your Productive plan exposes financials at a different path. An admin can fix it in Settings → Skills → productive_project_profitability → Financial Aggregation Path — no developer needed. (See "Good to know".)

2. Create the operations assistant​

  1. Go to the Agents area and choose Blank agent.
  2. Give it a Name (e.g. "Ops Assistant").
  3. Choose the Model Provider and Model (the recommended one is fine).
  4. In the Identity box, paste the Operations Assistant Identity below exactly — it tells the assistant how to behave for all three jobs.
  5. Under Tool Preferences, switch on the Productive tools — at minimum Project Profitability and Utilisation, plus the supporting List Projects and List People lookups.
  6. Click Save Changes.

Operations Assistant Identity (copy this into the Identity box)​

You are the Operations Assistant for our agency. You help the operations
team understand project profitability and team utilisation using our live
Productive data. You work in Australian English.

WHAT YOU DO
- Project profitability: when asked about margins, profit, or "which
  projects are losing money", call the Project Profitability tool. Present a
  table sorted least-profitable first, with revenue, cost, profit and
  margin % per project. Call out anything making a loss or below the margin
  the person mentions.
- Team utilisation: when asked "how busy is X", "are we overbooked", or for
  utilisation over a period, call the Utilisation tool. This reads the
  forward booking schedule (planned work), not past timesheets. Show each
  person's booked hours vs capacity and their utilisation %, a week-by-week
  breakdown across the window, and a team rollup. Flag anyone over 100%
  (overbooked) or under 50% (light).
- Finding the right people: for a role like "FEDs" / "front-end
  developers", pass the role so it matches on job title (e.g. "front"). If
  the person names a team or specific people, use those instead. If the
  result looks wrong or empty, say so and offer to try a team or named
  people rather than guessing.

PROACTIVE LOW-PROFIT FLAGGING
- When asked to watch for low-profit projects on a schedule, set up a
  recurring schedule. Use the margin threshold and timing the person gives
  you (e.g. "below 20%, every Monday 8am"). When the schedule runs, list
  only the projects below the threshold with margin, profit and a link. If
  none are below the threshold, send nothing. Post to the channel the
  person specified.

HOW YOU BEHAVE
- Always pull live figures from Productive at the moment of the request.
  Never invent, estimate, or reuse old numbers. If a tool returns nothing
  or errors, say so plainly and suggest the admin check the Productive
  integration's Test result in Settings → Integrations.
- You are read-only. You never create, edit, or delete anything in
  Productive — you only read and report.
- Utilisation % uses our configured weekly capacity (38 hours unless
  changed in settings). If someone questions the number, explain that
  capacity assumption.
- Money is in our Productive workspace currency. Don't convert it.
- Be concise and decision-useful: lead with the projects or people that
  need attention.
- If asked for a file, produce a clean document or spreadsheet in
  your organisation's brand (fonts and colours from the brand guide).

Want separate assistants instead of one? You can. Create a second assistant and trim the Identity to just the relevant job (e.g. a dedicated "Low-Profit Watcher" keeping only the Proactive low-profit flagging and How you behave sections). One assistant handles all three comfortably, so start there unless you have a reason to split.

3. Connect a channel​

Open the assistant → Channel Bindings, connect a Slack (or Teams) channel, and send a test message. This is where you'll chat with it — and where automatic flags will land.

4. (Optional) Tune the settings​

In Settings → Skills, two tools carry settings you can adjust without a developer:

  • Utilisation — Default Weekly Capacity (hours) (defaults to 38, Australian full-time), Default Window (weeks) (defaults to 4), and a safety page cap.
  • Project Profitability — Financial Aggregation Path (only change if the Test probe told you to).

You only do this setup once. After that, you just chat with the assistant in its channel.

Rolling it out and checking it works (do this first)​

Before the team relies on it, run these once. It takes about five minutes and confirms the two parts that depend on how our Productive workspace is set up.

Before deploying the update

  1. Build and run the tests so a release issue is caught before it ships:
    • pnpm --filter @enclaw/api build
    • pnpm --filter web build
    • pnpm --filter @enclaw/api test (the Productive tool tests run here)
  2. If all pass, deploy as normal.

After deploying — confirm the two live checks

  1. Test the Productive connection. Settings → Integrations → Productive → Test. In the Permission probes, you want a tick (✓) against Read Bookings and Read Financials.
    • If Read Financials shows a warning that the endpoint wasn't found, your plan exposes financials at a different path. Set it in Settings → Skills → Project Profitability → Financial Aggregation Path and Test again. (Common alternative: /financial_item_reports.)
  2. Run one profitability query. Ask the assistant "show me profitability for all projects" — you should get a table with real revenue/cost/margin. If it's empty, fix the path as above.
  3. Run one utilisation query and check the dates. Ask "utilisation for FEDs over the next 4 weeks" and confirm the week columns actually cover the next four weeks and the people look right. If the dates look off, tell your admin the bookings date filter needs checking; if the people look wrong, it's the job-title match (see "Good to know").

Once those three return sensible numbers, the team is good to go.

1. I want to know the profitability of our projects

Configure it​

Make sure Read Financials is ticked on the Productive integration and shows ✓ when you click Test. Nothing else is needed.

Use it​

Just ask, in plain language:

  • "Show me profitability for all our projects."
  • "Which projects have the thinnest margins right now?"
  • "What's the margin on the Acme website project?"
  • "Profitability for this financial year so far."

You'll get a table, least profitable first, with revenue, cost, profit, and margin % for each project — anything making a loss is flagged. Because the least profitable sit at the top, the projects that need attention are the first thing you see.

Handy variations

  • One project: name it and you'll get just that one.
  • A time window: "…between 1 April and 30 June."
  • A take-away file: "…and put it in a spreadsheet" or "…as a short Word summary" and it will produce a branded file you can share.

2. I want proactive flagging of low profitability

This is the "don't make me remember to check" one. You set it up once by asking the assistant to watch on a schedule; from then on it only speaks up when something needs you.

Configure it​

Nothing beyond having profitability working (section 1) and a channel connected. The threshold and timing live in plain words in the schedule you create — so you can change them yourself any time.

Use it​

Ask the assistant to set up a recurring check, and put your threshold in the request. For example:

"Every Monday at 8am, check project profitability. List any project with a margin below 20% — show its margin, profit and a link. If nothing is below 20%, don't message me. Post it to this channel."

That's it. Each Monday it pulls fresh figures and only messages you if a project is under your line.

Adjusting it later — just tell the assistant:

  • "Change the low-margin alert to 25%."
  • "Run the profitability check fortnightly instead of weekly."
  • "Pause the profitability alert." / "Turn the profitability alert back on."

Tip: keep the threshold where a breach is genuinely worth a look. If everything trips the alert, the alert stops being useful — raise the line.

3. I want to know our utilisation (e.g. FEDs over the next 4 weeks)

Utilisation here is forward-looking — it reads the booking schedule (planned work), not past timesheets. So "the next 4 weeks" means what people are booked to do, against what they could do.

Configure it​

Make sure Read Bookings is ticked and shows ✓ on Test. If your team's standard week isn't 38 hours, set Default Weekly Capacity in Settings → Skills → Utilisation.

Use it​

Ask for the group and the window:

  • "What's our utilisation for FEDs over the next 4 weeks?"
  • "Show utilisation for the design team for the next fortnight."
  • "Who's overbooked next month?"
  • "Utilisation for Jordan and Priya for the next 4 weeks."

You'll get each person's booked hours vs capacity and their utilisation %, a week-by-week breakdown across the window, and a team rollup. Anyone over 100% (overbooked) or under 50% (light) is flagged, so you can rebalance before it becomes a problem.

How it picks "FEDs" — it matches on people's job title in Productive (e.g. the word "front" finds front-end developers). You can also point it at a team or name specific people. If it returns the wrong group, see "Good to know".

What the % means — utilisation is booked hours ÷ available capacity over the window, counting business days only. Capacity uses your configured weekly hours (38 by default), so the percentage is only as accurate as that figure — adjust it in settings if your team differs.

Putting it together — a Monday operations rhythm

  1. Your low-profit flag fires (or stays quiet). If it names a project, you've got your starting point before you've finished your coffee.
  2. Ask for the profitability detail on that project to see how thin it really is.
  3. Ask for utilisation on the relevant team for the next few weeks to see who has room.
  4. Rebalance bookings in Productive — then next Monday the numbers reflect it.

No exports, no spreadsheets, no chasing people for their numbers.

Good to know

  • It's read-only. EnClaw reads Productive to answer you and to run the flag; it never edits projects, bookings, or budgets.
  • Figures are live. Every answer is pulled from Productive at the moment you ask (or when the schedule runs) — no stale copies.
  • Currency. Money is shown in your Productive workspace's currency.
  • "FEDs" relies on job titles. The group is matched on the job title in Productive. If your titles don't contain an obvious word (e.g. people are "Engineer II" rather than "Front-End Developer"), either name a team or list the people, or ask an admin to tidy titles in Productive.
  • Capacity is an assumption you control. Utilisation % uses a weekly-hours figure (38 by default). If part of the team is part-time or on a different standard week, the rollup is an approximation — adjust the default, or ask per-person.
  • If profitability comes back empty. That usually means the financial endpoint for your Productive plan differs from the default. Ask an admin to click Test on the Productive integration and follow the warning to set the Financial Aggregation Path in Settings → Skills.
  • Nothing is published without you. The assistant answers in your channel; the only thing it does on its own is post the low-margin flag you asked it to watch for.

Questions or something not matching what you see? Your EnClaw admin can check the Productive integration's Test results and the tool settings in Settings → Skills.