Runink TIDETIDE · Trusted Intelligence for Developer & Data Experience
Know what broke, what shipped and which number is right. On your servers, your cloud or ours.
Runink TIDE takes the repeat questions off your engineering, platform and data teams: why did it break, is the fix live, which dashboard is right, who changed this. Helpers draft the answer or the fix. A named person approves it. Run it on your own servers or cloud account and your data stays there. No plan sends it to an outside AI service.
A product in its own right, sold separately · in English, Spanish, French and Portuguese
Runink TIDEOne console, five parts. Each one takes a question off somebody's week.
- OverviewWhat is wrong right now, and where do I look first?
- DeliveryIs the platform up, and did my fix actually ship? (The console calls this part DevEx.)
- AI oversightWhich AI do we run, what may it do, and who checked its work? (The console calls this part DataEx.)
- IntelligenceWhich number is right, and where are money or controls slipping?
- FORGE (preview)How does a small tool a team asks for get built without waiting its turn?
Where does our information go? To the machines your plan names, and it stays there.
It says what it knows. It asks before it acts.
A dashboard that shows zero when it means I did not look gets the wrong person woken, or nobody. A tool that changes things on its own gets switched off after the first surprise. TIDE is built the other way round.
How the console is arranged →- Not known is not zero. A figure the console could not read is shown as not known, never drawn as a zero, so nobody decides on a reading nobody took.
- A person approves. The helpers propose. Nothing is filed, changed or published until someone with the right to decide says yes.
- A second opinion. Where a finding matters, an independent assessor reads the evidence first. It can agree, disagree or say it could not judge.
What it takes off your team's week
Each card starts with a burden your team carries. Then it says what TIDE does about it, and which cost it touches. A person approves anything that files, changes or publishes.
The call at 2 a.m. that a written reason could have answered
The person on duty has a screen for each system and none for all of them. Finding the problem takes longer than fixing it. TIDE opens on one Needs attention list, worst first. Each item has a plain reason and a link to the page that deals with it. Where TIDE has a remedy, it sits next to the reading, so the know-how belongs to the company, not to a few heads. A health check runs apart from the machines it watches, so it keeps watching when they fail. Cost it touches: specialist hours, and an on-call rota only a few names can staff.
Pull requests that wait days for a first read
A change sits until a busy senior engineer has time to look. TIDE’s reviewer helper reads changes on the repositories you choose. It lists what it found, worst first, with the evidence. Until a person arms it, it only reports what it would have said. Your engineers still decide what merges. Then Deploy lineage shows each change merged, reviewed, built and running, so is my fix live? needs no message to the platform team. Cost it touches: senior engineering hours spent on first reads and status questions.
Audit evidence rebuilt by hand before every review
Before each audit, somebody spends days piecing together who changed what, and on whose authority. TIDE writes it down as the work happens. Every console action is recorded with the person’s name, refused attempts included. Press Verify now on the Audit chain and it names the first record that was altered. One audit page gives a single verdict over the platform’s checks, and lists any check it could not read instead of counting it as a pass. Cost it touches: the person-days your last audit took, and outside help to prepare for the next one.
Paying for environments nobody remembers
A test environment is set up for a week and runs for months, because removing it is nobody’s job. In TIDE every deployment carries a named owner, a reason and an end date from the moment it is requested. It is flagged as the date gets close. When the date comes, the platform removes it. Cost it touches: the cloud or server bill for idle capacity.
Broken data, found by the business first
The first sign of a broken feed is often a manager asking why a report looks wrong. On one page, a data steward sees whether each source answered its last test, where a source’s own description disagrees with what it holds, and the open data-quality issues, and can check them all again on demand. Lineage traces each report back to the systems it comes from, with the trouble spots marked. Fixes wait in a queue where a person approves, tests or dismisses each one, and every decision is kept. Cost it touches: rework, and the hours spent tracing a wrong number back to its source.
Two teams, two numbers, one meeting
Finance and the programme office bring different figures and spend the meeting reconciling them. TIDE’s analyst, finance and programme-office dashboards are built on one set of figures, worked out the same way, from your own records. A figure TIDE could not read is shown as not known. Planned, requested and actual capital spending are checked against each other with a published rule book. Cost it touches: analyst hours spent reconciling spreadsheets, and money that leaves before anyone checks.
Nobody can list every AI helper in use
When the board, a customer or a regulator asks which AI you run and what it may do, the answer is a spreadsheet nobody trusts. TIDE keeps that list for you. It records each one’s source, version, licence and test evidence, checked against what is actually running, and what each helper is allowed to do. How much a helper may do on its own is a setting you choose for each kind of action, and every kind starts with a person approving each act. Cost it touches: risk and compliance hours spent answering the same questions again.
Data that cannot leave, so outside AI tools are off-limits
Customer records, contracts and claims cannot go to somebody else’s service, so the quick AI tools are ruled out, and promising evaluations end at where does that run? On the Dedicated and Enterprise plans, TIDE runs in your own cloud account or on your own servers. Its assistant, its helpers and FORGE run there too, and records, files, secrets and certificates stay there. On every plan, no outside AI service is called. Cost it touches: security reviews that stall a purchase, and a bill from an outside AI service that grows with every question.
No outbound connection, in our own test run
In our audit run, TIDE made no outbound connection while testing, exploring, mapping and assessing your sources; integrations you turn on, such as GitHub or Stripe, connect only to their own services. That was one run, about six seconds, on a development machine, with Chrome and the host out of scope, no model wired in, and no such integration configured for it.
The small tool a team needs waits behind the roadmap
In preview. A product owner describes the tool, or the scheduled steps that move and prepare a set of data, in plain words. TIDE proposes the steps on a canvas, and nothing is filed until that person approves. The approved brief goes to coding helpers as a work item. The result comes back as a change for your engineers to review. The brief stays on the machines your plan names; we are confirming there is no other path out before promising more than that. Cost it touches: engineering and contractor time spent on small internal tools.
Not billed by the question
Each new application rebuilds the same foundations: sign-in, secrets, a way to reach company data, somewhere to run, something watching it. Each tool bought to fill a gap brings its own account and its own bill. TIDE gives every application those foundations once. On every plan, heavy use of the assistant, the helpers or FORGE is not billed by the question. On your own servers, budgeting becomes a capacity decision made once, not a usage bill read every month. Cost it touches: engineering time spent rebuilding foundations, and servers you already own put to work.
See it on your own hardware
Before you start, write down two readings from your own records: the person-days your last audit took to answer who did what, and on whose authority, and every environment running today with the date it was last used. Both should fall once TIDE is in place. TIDE runs on one machine, from one downloaded file, with one command, and the version on your workstation is the version that runs on your servers.
In an afternoon
Open Overview and notice which readings say they have not been measured, instead of a reassuring zero. Open the findings queue (the console calls it the Harness) and read a proposed remedy. Open the Audit chain and press Verify now.
In a day
Add one real data source. Try to change it as somebody who is not on the permitted list, and read the record that refusal leaves. That is the record your security lead will ask for.
In a week
Put it on servers you own. Have somebody outside the platform team request a deployment, and watch it remove itself when its end date comes. Switch on the reviewer for one repository and read what it would have said before you arm it.
Read the detail
- The Runink TIDE paperEvery page of the console: the question it answers, who uses it, and why it is worth having.Read the paper →
- Runink TIDE and AtlasA joint paper with Logical Leap: Atlas's oversight screens inside TIDE, on your own data, with a second opinion built in.Read the paper →
- Trust & ComplianceWho decides, where the models run, the audit chain anyone signed in can check, and how our controls map to five standards. Each claim links to the public documentation behind it.Read the page →
- PricingPriced by where TIDE runs: per person on Runink's shared machines, as a team plan with an allowance of Compute Units; per machine in your own cloud; with you on your own premises. Sold on its own. No success fees, ever.See the plans →
Bring one system you would like to stop worrying about.
Half an hour, with whoever owns it in the room. We will show you what TIDE reads from it, what it keeps, and who can change it.
Data You Can Trust. Decisions You Can Defend.
runink.org/products/tide/