Security, Strategy And The Next Move: How Solution Providers Can Innovate In The AI Era
AI has left the hype stage. Training is giving way to "working models" that must produce returns. Security risk is growing, budgets are tight, and buyers want resilience-not more tools.
Here are the most useful takeaways from 10 tech leaders at XChange Best of Breed-and the moves you can make now.
Where AI Is Now: From Training To Working Models
Steven Carlini, Schneider Electric says we're past the first inning. With hundreds of large models in play and limited new public data, value shifts from training to inference and "reasoning" models that do actual work.
- Prioritize inference efficiency, latency, and cost per task. Measure $/ticket resolved, $/lead qualified, or $/document processed.
- Focus on retrieval, orchestration, and guardrails over building yet another base model.
- Track ROI per use case, not "AI spend." Kill or scale based on unit economics.
Malware At Scale: LLMs Lower The Barrier
Danny Jenkins, ThreatLocker points to a harsh shift: millions now have LLM access that can assist malware creation. The attack surface is larger and faster.
- Adopt default-deny: application allowlisting, privilege reduction, and script control.
- Assume social engineering succeeds. Segment aggressively and ringfence critical apps.
- Map controls to attacker techniques, not vendor buzzwords. See MITRE ATT&CK Initial Access patterns: MITRE ATT&CK.
AI Reality Check: Adoption Is Thin
Ayal Steinberg, IBM notes that while most have "initiatives," only ~25% have production projects, and fewer than 1% call their apps mature.
- Pick 2 revenue-facing use cases and move them to production in 90 days. Keep scope small, outcomes clear.
- Instrument every workflow: time-to-resolution, margin impact, error rate, customer NPS.
- Publish audited wins. Without proof, AI remains a cost center.
CX Matters: Simpler, Value-Based Resilience
Christian von Treskow, Commvault highlights a push to simpler architecture and competitive pricing while staying value-led.
- Price outcomes (RTO/RPO, immutability, recovery rehearsal), not SKUs. Reduce the "complexity tax."
- Offer small-footprint packages for SMB and standardized blueprints for mid-market.
- Make onboarding and policy design dead simple. Ease wins renewals.
Real ROI: AI That Lifts Margins And Throughput
Yolanda Cao, Everest Managed AI reports a client boosting margins by 11% without extra hires and triaging tickets 22x faster, then reselling the AI receptionist to their customers.
- Chase labor leverage and resellable assets. First, fix your own ops; second, productize for clients.
- Standardize metrics: margin delta, throughput per tech, lead volume, and burnout reduction.
- Create a resale catalog: AI receptionist, ticket triage, SOP copilots, and L1 deflection.
Project Management: Trust The Timeline Or Pay For It Later
Mike Psenka, Moovila points to a root cause: leaders still get "surprised" by projects. Unreliable timelines poison cash flow and customer trust.
- Adopt critical-path intelligence and automated risk scoring. Remove wishful dates.
- Gate work with ready-checks and dependency proofs before people start.
- Make forecast vs. actual a weekly ritual. Reward teams that call slips early.
Ransomware: Hygiene Still Decides Outcomes
Srikant Ganesh, Secureworks (a Sophos company) reminds us: weak passwords, no MFA, and stolen credentials open the door. The basics still drive most incidents.
- Enforce MFA everywhere, move to passkeys, rotate service creds, and audit stale accounts monthly.
- Immutable backups with clean-room recovery. Test restores like you test DR.
- Align to a clear framework to track coverage: NIST CSF.
Stop Selling "Cyber." Sell Business Resilience
John Pagliuca, N-able says customers don't want another point tool. They buy continuity and risk reduction.
- Frame offers around downtime avoided, financial exposure reduced, and compliance met.
- Bundle tiers by outcome (Good/Better/Best) with clear RTO/RPO and incident response commitments.
- Quantify impact: expected outage hours, cost/hour, and residual risk after controls.
Growth Outliers Are Bold (And Calculated)
Kevin Atkins, JumpCloud cites data: the top-growth MSPs take calculated risks.
- Set a 10% innovation budget with a 90-day prove-or-pivot rule.
- Place 2-3 bets: AI helpdesk deflection, identity-as-a-service upsell, and secure edge packages.
- Tie each bet to a P&L with leading indicators (pipeline, pilot success rate) and lagging ones (gross margin, churn).
Practical AI In ITSM, Without The Add-On Tax
Tim Barton-Wines, Halo notes their AI for ITSM is customizable and free inside the platform-rare in a field full of AI surcharges.
- Favor vendors who embed AI into workflows you already use and price fairly.
- Prioritize configurability over clever names. Control beats novelty.
- Score vendors on time-to-value: days, not quarters.
Your 90-Day Action Plan
- Pick two "working model" use cases: ticket triage and lead qualification. Ship to production with guardrails and cost tracking.
- Move to a default-deny posture: allowlisting, MFA everywhere, segmented access, and immutable backups. Test restores monthly.
- Repackage cybersecurity as resilience tiers with business outcomes, RTO/RPO, and board-level reporting.
- Fix project predictability: institute critical-path reviews, risk scoring, and weekly forecast vs. actual.
- Stand up an AI resale catalog: receptionist, SOP copilot, and L1 automation. Track margin lift per client.
- Fund 2-3 bold, time-boxed experiments. Kill fast, scale winners.
Upskill The Team (Optional)
If you need structured training to accelerate these moves, see role-based AI programs and certifications that map to delivery, sales, and ops:
The window is open. Shift from pilots to production, from tools to outcomes, and from caution to calculated bets. That's how solution providers win in the AI era.
Your membership also unlocks:
