GRAIL
Our work What we do Why GRAIL Essays Industry Briefings Book a call

Selected engagements

What changes when the leadership team sees it first

Scan the outcomes. Open any one to read what they needed, what we saw, and what we built, closing on the leader who lived it.

Jörgen Hedström, Pepperl+Fuchs
Pepperl+FuchsIndustrial sensors & explosion protection · family-owned · ~€930M

A skeptical sales team came around

Roughly 70% of a UK team that thought AI was noise turned in three workshops, and the pilot became the case for a rollout across the group.

The pressure

Pepperl+Fuchs is a ~€930M family-owned business, ~6,300 people across 40+ countries, competing against firms several times its size on engineering depth that lives in the heads of its most experienced people. The Group CEO backed a pilot on the hardest possible ground: the UK field-sales team, on live accounts, closest to the money. The question was blunt, does AI actually change what a thirty-year seller can do.

What we saw

In industrial B2B the product is physical and the advantage sits in the commercial layer, thousands of sellers whose judgment is the real asset. AI makes that judgment reachable for every seller, so a smaller company out-sells a larger one. We put the claim where it was most likely to fail, with a team that thought the whole idea was noise.

What we built, and how

Three workshops on their own live accounts, and a kit of 16 custom agents per seller, over 38,000 lines built on Pepperl+Fuchs’ own competitive landscape and sales motion, tuned between sessions from real use. A bar the leadership set in advance, capacity in weeks, pipeline in months, revenue in quarters. No systems touched, everything owned by P+F.

16custom agents38,000+lines3workshops0systems touched

The turn

A thirty-year veteran ran his kit more than a dozen times on the first day. Within a week, sellers were sending unprompted notes from the field about accounts the agents had surfaced.

Where it landed

By the final session a skeptical room had turned, and the pilot became the case for a group rollout, a geographic sales roll-out to ~350 sellers, marketing augmentation, sales-strategy advisory, and deep tracks for senior management and the board. We are in the roll-out phase now.

Jörgen Hedström
“What I like most is that you skill up our own people. The agents don’t replace our sellers, they make every one of them better.”
Jörgen Hedström · Pepperl+Fuchs ◇ ideal — confirming
Viktor Svensson, Rejlers
RejlersNordic engineering consultancy · Nasdaq-listed · ~5B SEK

A CEO-led AI programme, from the top

The group management team put AI on its own agenda, and the first tools GRAIL built paid for themselves against an outside vendor they replaced.

The pressure

Rejlers sells engineering expertise by the hour across the Nordics, Nasdaq-listed, ~5B SEK, ~3,400 people. Its Group CEO kept getting the same question from investors and his board, what is your AI position, and had no group answer. He framed it himself: in his own shoes he wanted three things proved, that AI shifts the key-account pipeline with a playbook that scales, that it takes real cost out of delivery, and that the management team can role-model the change from the front.

What we saw

An engineering consultancy’s economics rest on billable hours, and AI compresses the work those hours are sold for. The firm that repositions around outcomes early opens a lead its competitors cannot close for years. That call gets made in one room, the group management team, reasoning from their own hands on the tools.

What we built, and how

The Rejlers Group AI Lab, chaired by the CEO, on a monthly retainer with all IP to Rejlers from day one. Two tracks: KAM Intelligence, proving on five of the largest accounts before extending to every key account, and Elsikkerhet delivery modernization. Per-role agent kits including the CEO’s own. And Berik, a working tool that replaced an outside vendor and matched its output exactly, 5,400 cells, zero differences.

CEO-chaired AI Lab2tracksper-role kitsa vendor replaced
Top accounts
where we proved it
→
The division
a standing AI Lab
→
Whole motion
the sales org

The turn

“We have now tested Berik. And wow, everything correct on the first try.” The first tool GRAIL built had already paid for itself against the cost it removed.

Where it landed

The management team put AI on its agenda for the first time, and the programme kicks off in earnest with the Group CEO as active sponsor from August. Rejlers is building its own now.

Viktor Svensson
“It was the first hour this leadership team ever spent genuinely working out what AI does to our business. It changed the conversation.”
Viktor Svensson · President & CEO, Rejlers ◇ ideal — confirming
Mathias Berg, TV4
TV4Sweden’s largest commercial broadcaster · SEK 8.16B group

The leadership team got 63 working agents

Built against a contract that promised about 35, and weeks later the team had drafted its own list of AI bets for what comes next.

The pressure

TV4 had invested seriously in production AI, yet the people running the business worked much as they always had, while streaming, ad sales and audience data rarely met in the same room. With margins under pressure, leadership needed to become far more capable without going near the newsroom.

What we saw

Every media company pours AI into production while leaving the leadership team’s own judgment untouched, which is exactly where the coordination gap that costs them money lives. You can only set strategy for what you have experienced, so we made the leaders practitioners.

What we built, and how

The AI Enabled Workforce programme, three workshops (How to Think, Your AI Team, The Roadmap), a personal agent studio for every leader plus a layer for the CEO that pulls together the functional teams beneath him. 63 custom agents across ten leadership roles, all on TV4’s own context and owned outright. It closed on a roadmap of four AI bets, connect the systems, augment individuals, the sales cycle, finance reporting.

63agents10roles + CEO overlayowned by TV4
63agents
CEO overlaySalesFinanceStrategyContentCommercialMarketingPeopleProductData

The turn

Sixty-three agents against about thirty-five promised, and much of what the team had asked other vendors to build already sat built and unused.

Where it landed

Weeks after GRAIL left, the leadership team used the programme’s conclusions to draft its own AI Best Bets list and pulled GRAIL toward the next phase. The team now runs on tools built for its own work.

Mathias Berg
“You handed us the concrete agents, a way of working, and a real starting point for where AI takes TV4 next.”
Mathias Berg · CEO, TV4 ◇ ideal — confirming
Patrik Parnfors, Howwe
HowweB2B SaaS · strategy-execution software · deepest partnership

A consultancy’s work, built into the product

GRAIL owns and builds Howwe’s AI roadmap, turning what used to take a team of strategists into features that run for every customer at once. Nine co-developed AI features in the first phase, more coming.

The pressure

Howwe is one of the most AI-mature SaaS companies in Sweden. Its strategy-execution product lives on one hard truth, the average customer is not a strategy expert and cannot afford a management consultant to sit with every manager while a plan is laid and then watch every milestone in execution. That specialist is exactly what the product needed to become.

What we saw

In vertical SaaS the moat is the domain knowledge, and the win is turning it into features that let an average customer behave like best-in-class. AI can be the strategy specialist for each manager, connecting their goals to the broader strategy, breaking them into executable plans with real definition-of-done, and reading progress week over week the way a good consultant would.

What we built, and how

Working AI features co-developed with Howwe and handed to engineering as functioning prototypes to integrate and harden. Nine distinct customer-facing features including a Strategy Planner that turns a customer’s own documents into a board-ready plan, Strategic Pulse that reads each customer’s progress and reports the trajectory, initiative-evaluation models, a CEO growth-assessment app, and compliance drafting for tenders. Five are in production. Over 8,000 lines of working prototype handed to their engineering team.

9co-developed features8,000+lines of prototypea live roadmap
A strategist lays the plan with the manager
→
Strategy Planner turns their own documents into a board-ready plan
A consultant watches every milestone
→
Strategic Pulse reads each customer’s progress week over week
An analyst scores the initiatives
→
Initiative-evaluation models, plus a CEO growth-assessment app
An advisor drafts for the tender
→
Compliance drafting built into the product

The turn

The work that used to take a team of management consultants now runs inside the product, continuously, for every customer at the same time. No consulting firm can staff that. Software can.

Where it landed

An escalating monthly retainer through 2026 and a reseller partnership on top, with GRAIL continuing to own and drive the roadmap and advise the Howwe management team on AI.

Disclosure: GRAIL’s founder is a former Howwe CPO and shareholder.

Patrik Parnfors
“GRAIL turned the work that used to take a room of consultants into features our customers use every day.”
Patrik Parnfors · CEO, Howwe ◇ ideal — confirming
Christian Ask, Fitness24Seven
Fitness24SevenLow-cost 24/7 gyms · founder-owned · 280+ locations · 5 countries

46 agents for a leadership team, one per leader’s real work

We mapped 47 of their processes across 328 steps, then rebuilt each one around what AI now makes possible, and handed every leader a working kit.

The pressure

Fitness24Seven runs 280+ unstaffed 24/7 gyms across five countries on ~SEK 802M. Its founder saw that the constraint had moved off the gym floor and onto the hundred-plus-person central team, and that unless the management team itself stepped forward, an AI strategy would stay theoretical.

What we saw

Once the front line is automated the next gain comes from the knowledge work above it. The move was to make the leadership team builders, not users, and to never settle for doing the same job faster, augmented work means going further than the role could go before.

What we built, and how

A See The Opportunity series delivered in Spain and Sweden to the international management team. For each of nine leaders we captured how their work actually runs, 47 processes across 328 steps, understood the business logic behind each, then ran it through GRAIL OS to redesign it around what is now possible. 46 custom agents and over 17,000 lines written, each leader handed their own kit, their process maps, and skill stories teaching what is now automated, augmented, and newly possible.

46agents47processes328steps9leaders

The turn

The management team stopped asking how to do their own job faster and started asking what the role could become.

Where it landed

After the individual lens, the focus turned to the organization, where the grind lives and which roles have unutilized potential, feeding a strategy workshop on the bets that carry the highest ROI over the next twelve months. The team now has the experience to guide the company through what comes next.

Christian Ask
“My leadership team stopped asking how to do their jobs faster and started building things we could not do before.”
Christian Ask · Founder & CEO, Fitness24Seven ◇ ideal — confirming
Caroline Rudbeck, Infobric
InfobricNordic construction-technology software group · PE-backed (KKR) · ~350 staff

From four prompts to a decision engine on their own data

A marketing team went from four basic prompts to a nineteen-agent system, then a custom SEO engine that reads their live traffic and ranks the moves that actually shift pipeline.

The pressure

Infobric sells software to the construction industry across the Nordics and the UK, and its marketing team started from almost nothing on AI, four basic prompts, while a large pool of its own operating data sat gathered but unused, and a real budget went out to external agencies for SEO and web work.

What we saw

For a vertical software company the accumulated data is the moat, and it only pays off when it drives decisions. The fastest return was to read what already sat in their systems and rank the few moves that actually move pipeline, an orchestration layer on the substrate they already had.

What we built, and how

First, a marketing kit that grew to nineteen agents on their own brand and data, with an install manual and a verified design system. That earned a second phase, a custom SEO Optimizer built on their own dashboards, over 13,000 lines, a hardened data bridge and a crawler that read 1,572 pages of their live web, an engine that turns findings into a ranked list tied to the real revenue behind each gap, and the same pipeline producing the monthly marketing report senior management uses to steer spend.

19agent kitcustom SEO engine1,572pages crawledexec KPI report

The turn

The report that used to take a day by hand each month now writes itself, and the actions come ranked by the money riding on them.

Where it landed

The sponsor now works with GRAIL as an ongoing AI partner across the group, and the engagement grew from a first fixed-fee phase into a larger technical build.

Caroline Rudbeck
“It unlocked a whole new way of working. The data we were already sitting on finally tells us what to do next.”
Caroline Rudbeck · Group Head of Brand, Marketing & Communication, Infobric ◇ ideal — confirming
Kristin Gejrot, Coeli
CoeliIndependent wealth & asset manager · partner-owned · 70B+ SEK AUM

A day’s work, now an hour

Agents trained on Coeli’s own five editorial voices carry the production, so the team’s expert attention goes into shaping content rather than making it.

The pressure

Coeli manages more than 70 billion kronor and was named Sweden’s best wealth manager. Its marketing and investor-relations team was spending most of a day on a single newsletter, and wanted AI in that work without flattening the firm’s distinct editorial voice.

What we saw

In wealth management the scarce resource is expert attention, and much of the team’s was going into producing content rather than shaping it. Agents trained on Coeli’s own voice could carry the production and hand that attention back, at higher quality, with all five registers intact.

What we built, and how

Four workshops with the content team and a suite of custom agents on Coeli’s own brand voice and market context, all owned by Coeli, five voice profiles as reusable styles, a newsletter pipeline wired into their publishing, and research templates the team runs itself.

4workshops5voice profilesowned by Coeli

Where it landed

Newsletter production went from about eight hours to one, at higher quality, and the team uses the agents every day.

Kristin Gejrot
“The agents took over the production so my team could go back to the thinking. Same voice, a fraction of the time.”
Kristin Gejrot · Head of Investor Relations, Coeli ◇ ideal — confirming
A Nordic specialty-steel makerLong-products steel works · sales-led engagement

Turning a reactive sales floor into a forecast you can trust

Raw material runs about 80% of cost, and a thousand-ton forecast miss can swing revenue by tens of millions. We rebuilt the commercial process so the forecast becomes a control point.

The pressure

For this specialty-steel works, raw material is about eighty percent of total cost, exposed to nickel and currency, and a thousand-ton demand-forecast error can swing revenue by tens of millions. One recent quarter carried a multi-million loss traced straight to a forecast miss. The company had data in its systems, no defined sales process, and a culture that shrugged at the gap.

What we saw

The forecast was the business problem, and the forecast was only as good as a sales process nobody had defined. We framed the work as commercial development first, map how customers actually buy, attach KPIs and ownership, and turn the forecast into a control point, with AI compressing an engagement that would normally run for quarters into six weeks.

What we built, and how

A six-week sprint, an opening workshop, one-to-ones with each seller, and a kit of eleven named agents mirroring the firm’s own inquiry-to-quote process, from customer prioritisation and prospecting to meeting prep and quote reconciliation, with a prompting field guide and a roadmap, on a platform the client controls. Security sat at the gate throughout, and the sellers helped design the thing so it would actually get used.

11agents6week sprinttheir own processclient-controlled platform

Where it landed

The opening workshop delivered and the kit built against their real inquiry-to-quote process, with the final session after summer.

“For the first time the forecast is something we steer, not something that just happens to us.”
MD · Nordic specialty-steel maker ◇ ideal — confirming

Recently signed

Strawberry

Northern Europe’s largest hotel group. As AI moves in between the guest and the hotel across the discovery and booking funnel, we are building the leadership team’s strategy for where a hospitality brand sits once the funnel changes hands.

Delivery from August
Objektvision

Sweden’s commercial-property marketplace, part of a listed Nordic media group. A two-sided liquidity business whose real moat is the market data only it holds, and we are helping its leadership turn that data into the next products and pricing.

Delivery from August

Where do you start?

Start the conversation