Comparison updated
Side by side
Published data on both agencies. Where one side measurably differs, the dot marks it: the higher score, the lower published rate, the larger team, the more recent review. Which of those is an advantage depends on your brief.
| Attribute | Sonalake | Version 1 |
|---|---|---|
| Editorial score | 52.8/100 | Higher score: 61.8/100 |
| Hourly rate | Contact for rates | Contact for rates |
| Team size | 50-249 | Larger team: 250+ |
| Location | Dublin, Ireland | Dublin, Ireland |
| Founded | 2016 | 1996 |
| Last reviewed | Sep 3, 2026 | Sep 3, 2026 |
What we said about each
Our take on Sonalake
The one machine-learning case study with substance is the Vodafone congestion forecaster: cell KPIs decomposed into seasonal, trend and noise components, predicted with neural networks and time-series methods, then ranked so the first couple of hundred at-risk cells surface first—87% accuracy at the 97th percentile, 90% at the 99th. The rest of the AI evidence sits inside VisiMetrix, the firm's own network-analytics product, which has run at Three Ireland since 2015 and now carries ML…
Reviewed Sep 3, 2026
Read the full assessment →Our take on Version 1
The Foundation Risk Partners work is the study to read: an AI Labs system that reviews policy documents against the broker's business rules, hands the summary to a human expert and feeds the result back into the sales process, with the client's COO putting numbers on it himself—gap analysis from 40 hours to 4, policy checking that removed about $2. 5 million of outsourcing fees, and roughly $3. 9 million in revenue contribution so far.
Reviewed Sep 3, 2026
Read the full assessment →Best for
Where each agency measurably leads on the published data — team capacity, declared specializations, editorial scoring, and rates where both publish them. Use these to match an agency to your project priorities.
Choose Sonalake if you need
- closer collaboration at a smaller team scale
- disclosed specialization in Healthcare
- broader service offering — also covers AI Automation
Choose Version 1 if you need
- larger team capacity for multi-stream or enterprise-scale programs
- disclosed specialization in Public Sector
- broader service offering — also covers AI Consulting and Generative AI
- a higher overall editorial score
Strengths and watch-outs
Both upsides and risks, straight from our editorial assessments.
Sonalake
Strengths
- The congestion study publishes its method and separates prediction error from ranking error (70% and 88% respectively), which is rarer than a single headline accuracy figure
- Operator-level named references: a Chief Network Officer quote from Vodafone Ireland and a Head of Access and Transmission Networks quote from Three Ireland
- A named data scientist publishes an eight-step prototype-to-production process covering deployment, MLOps monitoring and model stability, so the delivery discipline is visible before the first call
Watch-outs
- The forecaster study reports accuracy, not business outcome—ask whether it is still in use at Vodafone and what it changed operationally
- Rates are unpublished; anchor on the €100–200/h EU band and ask whether AI work is staffed from Dublin or the Poznań engineering center
- The site names one data scientist and describes the wider organization as agile project teams—ask how many people do ML and who would be on your model work
Version 1
Strengths
- ISO/IEC 42001 certification for AI management announced January 2026, plus Claude Preferred Services Partner status from July 2026
- Named client executives speak on the record: FRP's CIO and COO, NAMA's Head of Systems, TNMOC's Museum Director and Barts Health's finance leadership
- Human-in-the-loop design is documented in two studies—expert validation before downstream content at FRP, per-field confidence scoring at Barts
Watch-outs
- FRP's headline figures (75% cycle-time cut, 300 bps EBITDA) are client-reported and cover a wider program—ask which parts Version 1 built versus advised on
- Treat the Department of Social Protection announcement as a 2026 framework contract award, not a delivery reference—no system is described yet
- Rates are unpublished; with staff across Dublin, London, New York, India and Australia, ask where your AI engineers would sit and at what blended rate
Service coverage
Where the two overlap, and where each covers ground the other does not.
Industry coverage
Where the two overlap, and where each covers ground the other does not.
Frequently asked questions
- Which has more FinTech experience, Sonalake or Version 1?
- Both agencies show documented FinTech work and have similar industry breadth. Compare directly on the agency profiles: Sonalake and Version 1.
- Which scores higher overall, Sonalake or Version 1?
- Version 1 scores 61.8/100, 9 points higher than Sonalake at 52.8/100. See our methodology for how scores are calculated.
- When should I consider both Sonalake and Version 1?
- Consider running parallel discovery briefs with Sonalake and Version 1 if your project spans multiple workstreams, you want competitive proposals to compare scope and approach, or you're undecided between the specialization angles each brings (see the Best for cards above). Most engagements ultimately go with one — but the parallel-brief phase is a low-cost way to validate fit.
Read the full reviews
A side-by-side is a starting point. The full assessments carry the portfolio analysis, documented engagements, and the evidence behind every strength and watch-out above.