TGH Tech
How we work

Start with one real gap. Prove it. Then expand.

Whether we’re building AI workflows, the platform your projects run on, or a founder’s first product, we work the same way: understand the work first, build in your name, put people in charge, and stay until it’s used.

[01]Our methodHow we work

Understand the work. Build the foundation. Fit the intelligence. Make it stick.

  1. 01Understand the work

    How it’s really done

    What comes in, who does what, where delays and errors happen, who approves, and what “finished” means. We sit with the person who does it best.

  2. 02Build the foundation

    Records and systems in order

    Data, integrations and platforms in your name, connected to the tools you already use. A clever tool on a shaky base just makes a mess faster.

  3. 03Fit the intelligence

    AI where it adds something

    AI for reading, judgment support and noticing. Fixed logic for anything that must be exact. A person confirming every result that matters.

  4. 04Make it stick

    Used, measured, handed over

    We build with the people who’ll use it, measure the difference against how the work was done before, and hand over the know-how.

[02]What every project has in commonHow we work

Four things you can count on.

  1. 01

    Scoped before we start

    You know what you’re getting and what isn’t included, in writing, before work begins.

  2. 02

    Built in your name

    Your methods, data, code and accounts are yours from the first day. We never resell hosting or software.

  3. 03

    People stay in charge

    Automate what can be safely defined. Assist where judgment helps. Keep human authority where responsibility demands it.

  4. 04

    Handed over with the know-how

    We teach as we go. The job is done when your people can run it without us.

[03]Measuring valueHow we work

We measure the difference in the work, not just the output.

Before we build, we record how the work is done today. Afterward, we test the same work the same way, including the review and correction it still needs.

What we measureBeforeAfter
SpeedTime and effort to finish the whole jobTotal time, including review and correction
QualityErrors, omissions and reworkError rate, exceptions caught, first-time acceptance
CapacityJobs done with the current teamJobs done at the same quality
TimelinessHow long problems go unnoticedTime from event to the right person acting
ControlUnclear ownership and missing sign-offShare of cases with correct evidence and sign-off
CostPeople’s time and the cost of mistakesTotal cost, including AI, review and support
ReuseDependence on one person’s knowledgeConsistency across people and repeat cases

A good demonstration shows what a workflow did. A good business case shows whether it produced a better result, on how many cases, at what quality and at what total cost. We aim for the second.

[04]How we keep it safeHow we work

We hold our own work to a written standard.

  • Rules the computer enforces: who can see what, how money is calculated, whether backups actually restore. Not left to memory.
  • Reviewed by someone else: nothing we build is reviewed only by the person who built it.
  • Anything that goes wrong twice gets an automatic check, so it can’t happen a third time.
  • Uncertainty is shown, not hidden: missing data, mismatches and failed connections are flagged instead of covered by a confident answer.
A written standard, enforcedOur standard
Who can see whatEnforced
How money is calculatedEnforced
Backups actually restoreTested
Reviewed by someone elseAlways
Wrong twiceAutomatic check

[05]What we don’t doHow we work

Said up front, so there are no surprises in month three.

We don’t decide what gets built. That’s always your call.
We don’t make professional judgments for you. Your engineers, accountants and managers decide. Our systems flag and route.
We don’t certify calculations or financial statements. Your named expert signs off, and the system records it.
We don’t do penetration testing or security certification. We’ll tell you when you need one.
We don’t give legal advice.
We don’t take work where the technology itself is the hard part, like very high-scale or real-time systems.

[06]More detailHow we work

Specific AI

Find the workflow. Make the intelligence specific. Measure the outcome.

The strongest first project isn’t a company-wide AI program. It’s one real workflow where generic AI isn’t enough, and where better execution can be shown.

  1. Step 1

    Identify

    Find a meaningful workflow: recurring work, avoidable cost or a missed follow-up that matters enough to fix.

  2. Step 2

    Understand

    Map the knowledge, context, rules and outcome: inputs, steps, who decides, the important exceptions, and what “done” means.

  3. Step 3

    Configure

    Combine models, data, tools and your specific logic on Bleno, with the review points and approvals you need.

  4. Step 4

    Deploy

    Test on real cases, then put it into real use, with your people confirming what matters and trained to run it.

  5. Step 5

    Improve

    Learn from reality, deepen it over time, and reuse what works for nearby workflows and other teams.

How a Specific AI engagement is structured

Your first workflow: Start with one workflow. We build it around your real work, you try it, and you only pay if it works. After that, an engagement has up to three parts, each agreed in writing:

  • Design and implementation (one-time): mapping the work, setting up Bleno, connecting your systems, testing and onboarding.
  • Bleno subscription (recurring): access to your configured workflows, agents and the platform.
  • Usage, support and expansion (as agreed): AI processing, support, new workflows and integrations.

A good first workflow has

  • Only one or two people can do it. When they’re busy or away, the work waits.
  • It follows clear rules. Checklists, rates, limits and standards.
  • It happens every week. Repeated work is where time adds up.
  • Mistakes cost money or trust. Rework, delays, unhappy clients.
  • Enough real examples to test on
  • An outcome you can measure
Own Your Stack

Software your business can own and depend on.

Specific AI needs good software underneath it. When nobody can vouch for that software, we make sure the business can: a review of its health and risks, safety rules for software your own team builds, deployment in your own name, and everyday tools you own instead of rent.

  • Start with one urgent problem: a software review, a deployment that depends on one person, or a tool you could own.
  • Put the right controls in place: releases, access, backups, code quality and who’s accountable.
  • Add regular reviews as your software grows.
  • Active support after handover.
For founders

Your first product, the same way.

  1. 01

    Foundation Sprint

    We learn your market, your users and your idea, and agree what to build first.

  2. 02

    Build

    A full team in your Slack and standups. Your first version is live in 8 to 12 weeks.

  3. 03

    Guide

    We help you decide what to build next and what to skip, including where Specific AI belongs in your product.

  4. 04

    Hand over, or keep going

    When you hire your own team, everything transfers to them.

Our rules

Seven rules that decide what we do on a normal Tuesday.

  1. 01

    A small business isn’t a small enterprise

    We work from what an owner-led business actually is: tight budgets, no IT team, and a crew that has watched tools get bought and abandoned.

  2. 02

    Trust comes before training

    Meet people where they already work, prove something early, and let them shape the system. People stand behind what they helped build.

  3. 03

    Foundations before AI

    Reliable records first, then written methods, then workflows, then agents.

  4. 04

    No AI for the sake of it

    Where plain software is enough, we use plain software, and we’ll say so.

  5. 05

    Out of the owner’s head

    How the business runs moves out of one person’s memory and into a system that works when they’re not in the room.

  6. 06

    We hand over the keys

    The job is done when you can run it without us.

  7. 07

    Weeks, not years

    You should see a working first result in weeks, on a real budget. If the first result is a year away, the plan is wrong.

Questions

The ones we hear every time.

Do we need an AI strategy first?

No. You need one real gap and the person who owns it.

Do we have to replace our software?

No. We connect to the systems you already use. Where a system is missing or can’t be trusted, we can build it.

What if we have nobody technical?

That’s most of our clients. You need one person who can make decisions about the work.

Will AI make decisions without us?

No. Some well-defined steps can be automated. Decisions about money, contracts or customers stay with a named person.

What does it cost?

Your first Specific AI workflow is scoped with you, and you only pay if it works. Own Your Stack work and founder products are scoped and priced before anything starts.

Who owns what you build?

You do. The methods, data, code and accounts are yours.

What happens when we want to stop?

You stop. Everything stays yours, with the documentation to carry on.

Start with one conversation and one real gap.

Which recurring piece of work would you most like to make faster, clearer or more consistent? Or what do you need built? Tell us, and we’ll tell you the first step and whether we’re the right people for it.