Rescue

Get to supported S/4HANA, with agents doing the heavy lifting

The path most ECC customers are already on, delivered by a small senior team and AI agents instead of a few hundred consultants. Same destination as your current plan, in less calendar time and for less money.

Who this is for

You have decided to stay with SAP, or the decision has been made for you. One of these is probably true:

  • ECC conversion is ahead of you. Mainstream maintenance for ECC ends at the end of 2027, with extended maintenance to 2030 at a premium.1 Most peers choose brownfield (44%), then bluefield (29%), then greenfield (26%).2 We run either of the first two. The assessment tells you which fits your estate.
  • A program has stalled or gone unstable. Scope has grown, the SI keeps asking for more people, and nobody can say with confidence when it ends. Large IT projects run 45% over budget on average.3 We can finish it or stabilize what is live.
  • You are on an older S/4HANA release. On-premise S/4HANA ships a major release every two years with seven years of mainstream maintenance.4 That means upgrades are a recurring cost, not a one-off.
  • You went live and it is not behaving. Jobs collide, interfaces back up, finance sees exceptions it cannot trace.

If you are not sure SAP-to-SAP is right, run the assessment first. It compares this path with Switch and Freedom on your own data.

The fastest route to supported status

If the deadline is what matters, do not try to perfect everything at once. Move the core financials and ledger as they are, and bring that onto S/4HANA first. Then handle the periphery (custom reports, interfaces, workflows around the ledger) with agents in the following waves.

Custom code is where programs slow down. Vendors estimate a quarter to a half of ECC custom code is unused, and systems commonly carry a few million lines.5 Agents read all of it, flag what is dead, and rewrite what remains for the target. A person reviews every change that touches money.

The people side is half the job. Staff who fly through SAP GUI transaction codes need retraining for Fiori. We plan that from week one, and agents help by drafting role-based guides from your actual process usage. Operating procedures that have drifted from how SAP really processes things are a common source of downstream errors that land on finance. The assessment finds them.

What agents do, and what people do

Agent shares are model assumptions, by hours of work. assumption

WorkstreamAgentsPeopleAgent share of hours
Custom-code cleanupInventory, usage analysis, dead-code flags, remediation for the target releaseReview of anything touching posting logic; sign-off on retirement90%
TestingGenerate and run regression cases from observed process behaviorDesign critical scenarios; run user acceptance cycles85%
Data migrationMapping, cleansing, trial loads, reconciliationDecide ownership and exceptions; finance sign-off on balances80%
Change managementDraft Fiori role guides, answer how-do-I questionsLead the organization through the change; train55%
CutoverRunbook drafting, rehearsal timing, checklist trackingRun the weekend; make go/no-go calls45%

Across the whole program, agents absorb about 74% of the work by hour. The calendar still has floors that AI does not compress: business decisions, user acceptance cycles, mock cutovers and period-end go-live.

How it runs

  1. Assessment (8 weeks for a large estate)

    Agents ingest configuration, custom code, usage, interfaces and document flows, and interview process owners. You get a semantic model of how the business runs, a conversion approach (brownfield or bluefield), and a fixed-fee quote.

  2. Core ledger first

    Convert the financial core as-is. Reconcile balances with agents; finance signs off.

  3. Periphery in parallel waves

    Agents remediate custom code, rebuild interfaces and generate tests, while the team's functional experts settle open design questions with your process owners.

  4. Rehearsals

    Mock cutovers against realistic volumes. Each rehearsal produces timings and a defect list for agents to clear before the next.

  5. Go-live and hypercare

    Agents watch incidents, batch runs and interfaces; people decide and act. Hypercare exits when agreed milestones are met.

After go-live: stabilization workflows

The same evidence layer that powers the migration keeps working afterward. SAP stays the system of record. Agents read approved evidence through narrow, read-only tools, and your SAP specialists validate causes and approve changes. Nothing writes to production on its own.

Incident timelines

Correlate errors, jobs, workload and transports into one ordered timeline, with source gaps stated and competing explanations ranked.

Batch and interface contention

Find overlapping jobs, queue growth and retries. Propose scheduling experiments for the quality system, each with a way back.

Finance exceptions

Compute exception counts and values from deterministic reports. Keep gross affected value separate from confirmed misstatement.

Change correlation

Line up incidents with transports and configuration changes. Timing is a lead, not proof; we ask for a reproducer before naming a cause.

Upgrade impact and test selection

Combine SAP's own check results with real usage and business criticality to propose a risk-weighted regression set. Mandatory control tests stay in, even for code that looks unused.

Release upgrades

Draft remediation work packages, rehearsal plans and cutover checklists for the next S/4HANA release, so the two-year cadence stops being a project each time.

SAP has announced its own migration assistants for early adopters, with no published effort reduction yet.6 They address a part of this work. We use whatever tools are available.

What it looks like at scale

Example: a company with $5โ€“20B in revenue, moderate customization, starting ECC conversion. Planning estimates from our model, not quotes. assumption

Delivery modelDelivery costCalendarTeam
Traditional SI, hourly (the current path)$24M26 moavg 35, peak 52
SI with AI copilots$20.4M24 mosimilar
Our small team, without agents$22.2M30 mo14
Our small team, with agents$11.8M12 mo14

The third row matters. A team of 14 senior people without agents would take longer than the SI does (30 mo against 26 mo) and save little. The agents are what make a small team viable. Our cost figure includes $2.3M of inference and 15% contingency.

For context, an indicative vendor-published benchmark puts brownfield programs planned at 15โ€“24 months at a median of 20โ€“30 months actual.7 Our figures assume no overrun on either side.

What you pay

Fixed fee, at our cost
$11.8M
no margin
Success fee (15% of current path)
$3.6M
only on acceptance, in tranches
Total
$16.2M
vs $24M on the current path
Difference
$7.8M
33% less, 14 mo sooner

The success fee is paid at milestones such as a passed rehearsal, go-live and hypercare exit. The run cost afterward stays at about $10M a year (assumption); this path does not change it. The assessment is a separate fixed fee at cost, about $755K for this size, and is useful even if you never hire us for delivery.

Risks on this path, and how we handle them

RiskWhat we do
Agents act on wrong or missing business contextThe assessment builds the semantic model first. Specialists review agent output on anything that posts to the ledger.
Estimates are optimisticEvery figure is a planning estimate. The fixed fee is set after assessment, with contingency in it, and the success fee only pays on acceptance.
Business decisions and UAT set the paceCalendar floors are built into the plan; agents cannot compress them. We ask for decisions early.
Users resist the move to FioriRetraining and procedure cleanup start in the first month, not the last.
Production changes made by automationAgent tools are read-only in production. People authorize and execute changes.

Sources

  1. The Register, 28 Jan 2025. SAP extends support deadline.
  2. ASUG, 3 Oct 2025. SAP S/4HANA adoption trends (168 members).
  3. McKinsey & University of Oxford, 2012. Delivering large-scale IT projects.
  4. SAP, 2022. New SAP S/4HANA release and maintenance strategy.
  5. SAP Community, Aug 2026. ABAP custom code migration process. The unused-code range is a vendor estimate, not SAP-published.
  6. SAP News Center, 28 Oct 2026. Migration and Modernization Assistants.
  7. VendorBenchmark, Apr 2026. S/4HANA migration cost benchmark. Indicative vendor-published benchmark.