Skip to content

Pilot

A three-week pilot on one flow you choose.

The pilot answers one question: can you trust Telikos to explain this part of your estate, with proof, faster than your team can today?

One flow, not the whole estate

We agree one painful flow with you, such as nightly claims adjudication or month-end close: the jobs, programs and data that make it up, and three to five questions your team struggles to answer today. The questions go into the pilot charter word for word.

Preparation takes one to two weeks and runs alongside contracting: your security review, a host, network rules, read-only access to the source, and the choice between a local model and none.

The three weeks

  1. Week 1Secure access and ingestionWe install the appliance on your host and create accounts for your team. We connect read-only to your source, ingest the flow and check parser coverage. With your infrastructure contact we walk through the unknowns, ingest the missing artifacts, and show your pilot lead the estate in maps, a first cited answer and the gaps that remain.
  2. Week 2Joint review with your expertsIn two or three short sessions your experts work through the review queue. They confirm, correct or reject the summaries and rules for the flow's key programs, and record what the code cannot show: dynamic calls, manual steps, owners and schedules. We ingest again and confirm that every correction survives.
  3. Week 3Your team asks real questionsYour team asks the charter questions word for word. For each answer an expert follows the citations to the source and grades it. We run the impact question for a real pending change, trace a field you care about, and time the same questions the old way where you can.
  4. Close-outA report and a decisionA final report covers the success criteria, measurements, graded answers, open gaps and what your experts added. Then you decide: subscribe, extend or stop. If you stop, everything is removed from the host and you confirm the deletion in writing.

What you keep

  • The appliance, running on your host, with your flow ingested and backed up.
  • Maps of the flow across its jobs, programs, datasets and tables.
  • The unknowns for the flow before and after, with what closed each gap.
  • Facts and rules for the flow's key programs, reviewed by your experts.
  • Cited answers to your charter questions, graded by your experts and exported as Markdown or PDF.
  • The final report, with the measurements behind each success criterion.

How we judge it

You agree the success criteria with us before week 1, using measurements the product records. These are the thresholds most pilots start from.

Pilot success criteria
CriterionMeasured byTypical threshold
Parser coverage on the flowThe ingestion summary95% or more
Unknowns on the flowThe unknowns before and afterEvery one closed or explained
Charter questions answered correctlyYour experts' grades in week 3At least 4 of 5
Unsupported claims shown as factsEach answer's count of statements dropped for lack of evidenceZero
Expert corrections that survive re-ingestionThe re-ingestion in week 2All of them
Time to answer, compared with the old wayWeek 3 timingsFaster on most questions

What we need from you

Artifacts
The source for the flow: COBOL programs, copybooks, JCL and procedures, DB2 DDL, and CICS definitions if the flow is online. No production data.
Read-only access
A Git deploy token or SSH key, an SFTP account, a read-only share mounted on the appliance host, or an agreed zip export, delivered through your own secret channel.
A host
8 vCPU, 32 GB of memory and 100 GB of disk with Docker and disk encryption. Add a GPU with 16 GB or more if the pilot uses a local model; without one, answers are composed from evidence alone.
Network rules
Port 443 reachable from your experts' desks, and the appliance able to reach your Git or SFTP server or the share it reads from. Nothing else.
People
A pilot lead who owns scope and success criteria, an infrastructure contact, a security reviewer, and two to four subject-matter experts for a few short sessions.

Pricing

The pilot is a fixed fee, set by the size of the flow, and credited in full toward the first year's subscription when you sign soon after it.

Subscription pricing is scoped with you from what the pilot showed: the systems in scope and how many people will use it. Annual, prepaid. Expansion pricing is written into the contract, so adding a system never needs a re-quote.

We quote in conversation, not on a page. Tell us about the flow and we will tell you plainly.

Request a pilot

Tell us about the flow and how the source could be shared. We will reply to set up a first conversation with your pilot lead.

For example, nightly claims adjudication, or the jobs behind month-end close.

How could the source artifacts be shared, read-only?

We use what you send only to reply to you about a pilot.

Prefer to talk first?

Book a conversation with the people who build Telikos, or email us with your questions.