GoTo: How AI Is Changing the Future of Work
AI is no longer a side project. It sits inside the modern workplace stack, influencing productivity, security and how teams communicate day to day. GoTo, a leader in flexible workplace communication and IT management, is building for a hybrid-first reality where user experience meets enterprise-grade resilience.
On 18 November 2025 at Tech & AI LIVE New York, Olga Lagunova, Chief Product & Technology Officer at GoTo, will share a clear view on the hype vs reality of AI - with a focus on what actually works in production.
Building tech for the modern workforce
Under Olga's leadership, GoTo prioritises scalable, secure platforms that let organisations connect, support and manage teams from anywhere. The company's flagship products, GoTo Resolve and GoTo Connect, apply intelligent automation and AI-driven analytics to streamline IT operations and improve end-user outcomes.
The goal is simple: reduce friction for distributed enterprises without adding complexity. That means smarter routing, faster resolution, tighter feedback loops and a clear path to measurable results.
Responsible AI as a product principle
GoTo's approach is human-first. Automation should augment critical judgment, not replace it. Predictive insights, workflow optimisation and AI-enabled support help teams perform with confidence while keeping security front and centre.
- Human-in-the-loop by default: Pair AI suggestions with clear controls and escalation paths.
- Transparent outcomes: Explain the "why" behind actions, not just the "what."
- Guardrails: Data minimisation, PII redaction and role-based access as non-negotiables.
- Measurement: Track business impact, not just model metrics.
Hype vs reality: what product leaders should focus on
Pilots are easy. Production is where it gets real. Olga's session will focus on scaling responsibly, managing risk and converting promising prototypes into durable systems that customers trust.
- Start with the job to be done: Prioritise high-friction workflows with clear success metrics.
- Data readiness over model novelty: Clean pipelines beat fancy models with messy inputs.
- Evaluation that mirrors reality: Golden sets, scenario testing and live A/Bs, not just offline scores.
- Observability: Log prompts, responses, latency, cost and drift. Alert on failure modes early.
- Controlled rollouts: Shadow mode, feature flags and phased exposure to manage risk.
- Safety and compliance: Apply policy checks, content filters and access controls from day one.
From pilot to production: a simple playbook
- Discovery: Define the business case, target KPI and user story. Document failure modes and guardrails.
- Prototype: Build a thin slice with real data. Add human review. Prove value on a narrow task.
- Pre-Production: Add eval suites, red teaming, privacy review and audit logging. Run in shadow.
- Launch: Feature-flag to a small cohort. Monitor quality, cost per action and user feedback daily.
- Scale: Automate retraining or prompt updates. Set SLOs. Establish an on-call runbook for incidents.
Practical patterns for product teams
- Right-size the model: Use smaller models for routine work; reserve larger models for complex cases.
- Grounding and retrieval: Keep answers consistent with your knowledge base. Cache frequent results.
- Tool use with constraints: Define allowed actions and strict input/output schemas.
- Fallback logic: If AI confidence is low, route to rules or a human without delay.
- Feedback loops: Capture user corrections and close the loop in updates or fine-tunes.
Sustainable innovation that pays for itself
GoTo's roadmap emphasises efficiency that compounds. That includes reducing resource waste, improving service reliability and simplifying operations at scale. Track unit economics like cost per ticket resolved, mean time to resolution and emissions per request.
- Cost controls: Token limits, response truncation, batching and cache hit targets.
- Reliability: Deterministic fallbacks, retries with backoff and structured output validation.
- Resource use: Prefer retrieval + small models before calling larger ones. Monitor GPU minutes like a budget line.
Security and governance without the drag
Trust is a product feature. Bake in governance so your team moves faster, not slower.
- Data policy: Minimise data, mask PII, separate environments and expire logs on a schedule.
- Vendor risk: Review model providers, region controls, subprocessor lists and SLAs.
- Model accountability: Document purpose, data sources, eval results and known limits.
- Standards: Consider the NIST AI Risk Management Framework and monitor guidance under the EU AI Act.
Team structure and skills that ship
High-output teams pair product, design and engineering with data, security and support from the start. Define owner roles for prompts, evaluations, data pipelines and incident response. Create space for experiments, then promote what works into the core roadmap.
If you need a faster way to upskill product and engineering teams on AI practices, explore curated paths by role at Complete AI Training.
Tech & AI LIVE New York 2025
Tech & AI LIVE New York runs on 18 November 2025 as a virtual conference for leaders building the next wave of digital work. Expect keynotes, panels and practical sessions across AI adoption, digital collaboration and cybersecurity.
Olga will join a strong lineup including Joe Depa (EY), Kevin Miller (IFS) and Venky Santhirahasan (PepsiCo). To attend, look up Tech & AI LIVE New York and register for free virtual tickets.
The bottom line for product development
AI is useful when it reduces friction you can measure. Focus on problem fit, clean data, clear guardrails and staged rollouts. Keep humans in the loop, watch cost and reliability like product KPIs and build a feedback system that gets smarter over time.
That's how AI moves from buzz to business results.
Your membership also unlocks:
