Five ways to work with us
Prices are shown for Indiathe United Statesthe United KingdomEurope — change region. Every engagement starts with one process, and you can stop after any step and keep everything we delivered.
Prices in
The audit
- Who it's for
- Operations leads at companies of 20 to 1,000 people whose teams run on a mix of software, spreadsheets and email, and who suspect a few processes cost far more than they should.
- Who it isn't for
- Teams with nothing running yet. There's no process to map.
Two weeks mapping how the work actually runs, which parts are worth automating with the time and cost against each, and which parts to leave alone. You keep the report whether or not you hire us. An audit that only pays off when it recommends more work from the same firm is a sales pitch with a process map attached.
Deliverables
- A map of each process as it runs today, built from your systems and your team's own account of them
- A ranked list of candidates, each with the hours it takes now and an estimate to build and run
- A written list of what not to automate, with the reason for each
- The integration risks we found, such as rate limits, duplicate records and documents that change format
- A recommended order of work
What you can do afterwards
- You can brief any engineering team from the report, including one that isn't us
- You know which processes should never see a model
- You can price a build before you commission it
Not included
- Writing or deploying code (that's a pilot)
- Negotiating with vendors
- Advice after the report walkthrough
- How long
- Two weeks from kickoff. On-site days are available, with travel billed at cost.
- Price
- from ₹45,000from $3,500from £2,500from €3,000
Credited in full against a pilot started within 90 days.
Private AI readiness check
- Who it's for
- Teams that want AI on their own machines or in their own cloud and need to know what their hardware can handle.
We install a small benchmark on your machines, run it on examples from your own work, and tell you what each machine can run, for how many people at once, and where you'd need something bigger.
Deliverables
- An inventory of the machines we tested
- Speed and concurrency results for each workload you care about — document reading, drafting, chat, speech
- Accuracy on 20 to 50 of your own examples, compared across two or three models
- A plain verdict per workload and a recommendation — your machines, an India-hosted cloud account, or a partner machine — with costs
What you can do afterwards
- You know what to buy, or that you don't need to buy anything
- You can hold any vendor's performance claims up against your own numbers
Not included
- Hardware (a partner invoices you directly)
- Building the tool itself
- How long
- About one week.
- Price
- from ₹75,000from $5,000from £4,000from €4,500
The pilot
- Who it's for
- Teams that know which process is costing them and can name the systems it touches, usually after an audit.
- Who it isn't for
- Anyone who wants a general assistant across the whole company. One pilot, one process.
One tool, built around one process, deployed in your environment in four to six weeks. Your keys, your audit log, the full source in your repository, and a runbook written for whoever is on call.
Deliverables
- One tool deployed inside your environment, working against your systems
- Full source in your repository, under your licence, with its commit history
- An audit log, as a table in your own database
- A runbook — what can go wrong, what it looks like, what to do
- An evaluation suite and dataset you can re-run yourself
- A recorded handover
What you can do afterwards
- You can run, change and redeploy it without us
- You can show an auditor who did what, from a log we can't write to after handover
- You can test a new model version on your own examples before you switch
Not included
- Your cloud, hardware and model costs
- Third-party licences, bought in your name
- Changes to the systems on either side
- 24/7 on-call
- How long
- Four to six weeks, depending on how many systems it has to reach.
- Price
- from ₹4 lakhfrom $15,000from £10,000from €12,000
Enablement
- Who it's for
- The people who'll use or run what we build, and teams starting to build with AI themselves.
- Who it isn't for
- Awareness sessions for hundreds of people at once.
Hands-on sessions built from your own processes and documents, in small groups, with assessments on cases nobody in the room has seen before.
Deliverables
- Sessions around your processes and systems
- Written material your team keeps, with worked examples
- Assessments on unseen cases, scored and returned
- A short read-out on where the group was strong and where it wasn't
What you can do afterwards
- Your team can tell which problems need a model and which need a rule
- The people on call can read an audit log without help
- You have a record of who was trained on what
Not included
- Certification
- Recorded courses or a learning platform
- Travel (billed at cost)
- How long
- Half-day or full-day blocks, usually over two weeks.
- Price
- from ₹1 lakh a dayfrom $4,000 a dayfrom £1,500 a dayfrom €1,800 a day
Run and improve
- Who it's for
- Teams with a Stafflo tool in production.
Monthly support with a named contact, fixes, and model updates tested on your own examples before they reach your staff.
Deliverables
- A response within one working day, during business hours in India
- Model and dependency updates, each tested against your evaluation set first
- A monthly note on what changed and what the tests showed
What you can do afterwards
- A model update can't quietly make your tool worse, because it doesn't ship until it passes your tests
Not included
- 24/7 on-call
- New tools (each one is a pilot)
- How long
- Monthly. Cancel with 30 days' notice.
- Price
- from ₹40,000 a monthfrom $2,000 a monthfrom £1,500 a monthfrom €1,800 a month
Support and response
We reply within one working day, during business hours in India, with a named escalation contact from kickoff. We don't run a 24/7 on-call rota.
What we won't claim
We don't quote results for anyone's revenue, conversion or headcount, and we have no client results to quote yet. What we do show is what each tool does and what our own tests measured, with the run behind every number.
Questions people ask before they book
Do you need access to our data?
For a pilot, yes, inside your environment. We work in your systems through an account you create and can revoke at any time. Nothing is copied to us, and the audit log records everything we do.
What if the model gets worse after an update?
Under Run and improve, every update is tested on your own examples before it ships. If it scores lower, it doesn't ship, and the monthly note says why.
You're a small studio. What happens if you disappear?
You own the source, the runbook and the tests from the first day, in your own repository. Any competent team can take the tool over, and our refund terms cover any work we can't finish.